IndyDevDan
Lets Get Fired: Using AI Coding Assistant AIDER to do my Engineering Job
updated
🎥 Featured Media:
Part 1: Realtime API with tool chaining
youtu.be/vN0t-kcPOXo
Part 2: Senior Engineer’s AI Assistant
youtu.be/090oR--s__8
Two way prompt video
youtu.be/sTruFeIO0iA
🚀 In this video, we breakdown the 2025 plan to supercharge your AI engineering skills. We'll explore how advancements like Sonnet 3.5, O1 reasoning models, structured outputs, and the realtime API are changing how we build with AI. This is about you wielding as much compute as you can use build AI Agents and Assistants.
🔥 We introduce Ada v3, a personal AI assistant built for engineers. Watch Ada effortlessly navigate databases, generate SQL queries, create documentation, and even build Python charts. This is a glimpse into the future of agentic engineering, where AI handles the heavy lifting, freeing you to focus on high-level design and strategy.
🛠️ This video isn't just about tools; it's about a mindset shift. We break down the core components of modern AI engineering: prompt design, AI agents, AI assistants, and ultimately, agentic workflows. Learn how to leverage these components to maximize your AI compute and achieve unprecedented productivity gains. We'll cover everything from prompt engineering and prompt design to building powerful AI agents and personal AI assistants.
🌟 This video is essential for any AI engineer, software developer, or tech enthusiast looking to stay ahead of the curve. Learn how to harness the power of AI compute, master agentic engineering principles, and transform your workflow. Whether you're working with Sonnet 3.5 or other LLMs, this video will equip you with the knowledge and strategies to succeed in the age of AI.
💡 Key takeaways:
AI Assistant: Build your own personal AI assistant like Ada.
AI Agents: Learn how to design and deploy powerful AI agents.
Reasoning Model: Understand the role of reasoning models like O1.
AI Compute: Maximize your AI compute for optimal performance.
Prompt Engineering: Master the art of prompt design for effective AI interaction.
Agentic Engineering: Embrace the principles of agentic engineering for next-level productivity.
Realtime API: Leverage the power of realtime APIs for seamless AI integration.
Join the journey.
Stay focused, and Keep building.
📖 Chapters
00:00 Four key breakthroughs for 2025
00:44 Personal AI Assistant Ada v3
03:05 My 2025 Plan For AI Engineering
17:27 Three Big ideas for the rest of 2024
#aiengineer #aiagents #aicomputing
What if your AI Assistant could WORK in parallel to you?
💻 Get Your Assistant (CODEBASE)
github.com/disler/poc-realtime-ai-assistant
🎥 Featured Media:
- Watch Part 1: youtu.be/vN0t-kcPOXo
- OpenAI Structured Outputs: platform.openai.com/docs/guides/structured-outputs/introduction
- OpenAI Swarm: github.com/openai/swarm
- OpenAI Pricing: openai.com/api/pricing
🔥 Watch as we demo a cutting-edge AI Assistant that's about to revolutionize how engineers work. No typing, just speech to speech interactions - and getting things DONE!
In this video, we showcase the power of AI assistants in accelerating information processing and manipulation - the core of software engineering. Watch as we effortlessly:
🚀 Scrape and clean web content
🖥️ Generate and run Python code
📊 Create and modify CSV files
🧠 Utilize active memory for enhanced context
Witness firsthand how AI assistants can work in parallel with you, handling tasks while you focus on high-level thinking. We'll explore:
1. The game-changing combination of reasoning models and real-time speech-to-speech APIs
2. The importance of active memory in AI assistants
3. How specialized AI agents can be delegated tasks through natural language
This isn't just about automating mundane tasks - it's about supercharging your productivity and creativity as an engineer. Imagine having a tireless assistant that understands context, learns from your work patterns, and executes complex tasks with minimal input.
💡 Key Takeaways:
- The future of software engineering lies in parallel processing with AI assistants
- Active memory management is crucial for effective AI assistance
- Natural language control of specialized AI agents is becoming a reality
- This technology is set to dramatically change how we approach building software
Whether you're a seasoned developer or just starting out, this video offers a glimpse into the future of software engineering. Don't miss out on this opportunity to stay ahead of the curve!
Like, subscribe, and share your thoughts on how AI assistants could revolutionize your workflow. The future of software engineering is here - are you ready?
📖 Chapters
00:00 Show not tell
00:25 AI Assistant Engineering
03:30 Speech to Speech Learning
06:50 File and Data Manipulation
09:33 Engineer's AI Assistant Discussion
#aiassistant #aicoding #promptengineering
🤖🔥 ADA is BACK!
It's time to start rethinking how we interact with AI assistants. ESPECIALLY for software engineers.
🎥 Featured Links:
- Python Async Realtime API POC codebase:
github.com/disler/poc-realtime-ai-assistant
- OpenAI Realtime API:
openai.com/index/introducing-the-realtime-api
- Super AI Agents with Structured Outputs:
youtu.be/PoO7Zjsvx0k
- Control Your Personal AI Assistant:
youtu.be/ikaKpfUOb0U
- One Prompt is Not Enough:
youtu.be/JjVvYDPVrAQ
🔥 In this game-changing video, we're unleashing the full potential of personal AI assistants like Ada. Discover how the new OpenAI Realtime API is tearing down the barriers between you and your digital assistant, enabling real-time tool chaining and function chaining like never before!
🛠️ See firsthand how Ada utilizes the o1 assistant and advanced AI agents to perform complex tasks with 100% accuracy. We'll dive deep into the mechanics of tool chaining and function chaining, showcasing how these techniques can transform your interaction with your personal AI assistant.
💡 Whether you're an engineer, developer, or AI enthusiast, understanding these cutting-edge techniques is crucial in the age of AI. We'll explore the trade-offs, discuss the risks, and explain why embracing the OpenAI Realtime API is worth it for engineers who want to stay ahead.
🚀 I, IndyDevDan, break down complex concepts into easy-to-understand insights. From experimenting with file AI agents to implementing personal AI assistant patterns, we'll guide you step-by-step through the revolutionary capabilities of Ada powered by the OpenAI Realtime API.
🌟 Don't forget to like and subscribe for more edge content on AI, automation, and the future of personal assistants!
📖 Chapters
00:00 ADA is back.
00:49 OpenAI Realtime API
03:02 o1 File CRUD AI Agent
06:08 Breaking down tool chaining
06:34 Experimenting with file ai agent
10:03 Personal AI Assistant Patterns
12:25 Realtime API Tradeoffs
13:22 It's worth the risk, engineers NEED this.
15:45 Wow, 20k subs soon, our focus has not changed.
#promptengineering #aiassistant #programming
🔗 Resources:
- Multi LLM + SLM Codebase: github.com/disler/marimo-prompt-library
- Meta Llama 3.2: ai.meta.com/blog/llama-3-2-connect-2024-vision-edge-mobile-devices
- Ollama: ollama.com
- Marimo: marimo.io
🔥 Small Language Models (SLMs) are heating up
In this video, we dive deep into Meta's Llama 3.2 3B and 1B parameter models and evaluate whether small language models are ready to rival the big players in the LLM arena. Using Ollama and Marimo, we compare the performance of Llama 3.2 against models like GPT-4o-mini, Sonnet, Qwen, Phi, and Gemini Flash. Are SLMs like Llama 3.2 finally good enough for your projects? Let's find out!
🔍 Hands-On Comparisons Beat Benchmarks Any Day!
We run multiple prompts across multiple models, showcasing real-world tests that go beyond synthetic benchmarks. From code generation to natural language processing, see how Llama 3.2 stacks up. Discover the surprising capabilities of small language models and how they might just be the game-changer you've been waiting for.
🛠 Tools to Empower Your AI Journey
We'll also explore how tools like Ollama and Marimo make it easier than ever to experiment with small language models on your local device. Whether you're into prompt testing, benchmarks, or prompt ranking, these tools are essential for maximizing your AI projects and understanding what small language models can do for you.
Join us as we uncover whether SLMs like Llama 3.2 are truly ready to take on the giants of the LLM world. If you've been curious about the latest in prompt testing, benchmarks, and prompt ranking, this is the video for you!
📖 Chapters
00:00 Small Language Models are getting better
00:40 How good is llama 3.2 REALLY?
01:17 Multiple Prompts on Multiple Models
08:32 Phi, Llama, Qwen, Sonnet, Gemini Flash model voting
13:53 Hands on comparisons beat Benchmarks anyday
18:38 SLMs are good, not great but they are getting there
#promptengineering #softwareengineer #aiengineering
✅ Marimo Reactive Notebooks
marimo.io
🔗 Build Your Reusable, Interactive Prompt Library
github.com/disler/marimo-prompt-library
💡 "We Were right" - o1 Reasoning Mode
youtu.be/GUVrOa4V8iE
💻 Master AI Coding
youtu.be/ag-KxYS8Vuw
If it's not clear, the prompt is everything. Your ability to create, reuse and iterate on your prompts may be THE deciding factor in your success as an engineer in the age of LLMs and generative AI.
In this video I have 3 course special that can help you build your personal prompt library.
It all starts with Marimo.
In this video, we introduce Marimo, the next-gen Python notebook designed to replace traditional Jupyter Notebooks. Discover how Marimo Reactive Notebooks empower you to reuse, iterate, and visualize your prompts effortlessly. Whether you're using o1-mini, o1-preview, Claude, or Gemini, Marimo provides the perfect platform for prompt design and agentic engineering.
Learn how to build and maintain a robust prompt library with reusable prompt templates and prompt variables, enhancing your productivity as an engineer in the generative AI age. Marimo's interactive features allow you to run individual prompts across multiple large language models with just a click, making prompt engineering more efficient and effective than ever before but that's just one idea. It's a python notebook, so you can build ANYTHING YOU CAN IMAGINE.
From rapid prototyping to interactive data visualization, Marimo Notebook is your ultimate tool for mastering AI coding. Join us as we dive into the features that make Marimo a game-changer for engineers looking to stay ahead in the age of generative AI. Don't miss out on building a powerful and scalable prompt library that will elevate your AI projects to the next level!
📖 Chapters
00:00 Introducing Marimo - Next-Gen Python Notebooks
01:15 Goodbye Jupyter Notebooks, Marimo First Look
02:13 Marimo Reactive Notebooks - Reuse, Iterate, Visualize
06:25 Instant Prompt Notebook with o1, Claude, Gemini
14:22 Engineer Your Prompt Library with Marimo
24:11 Marimo - Version Control and AI Codable
#marimo #promptengineering #agentic
📚 Resources:
- Simon Willison's article: simonwillison.net/2024/Sep/12/openai-o1
- Simon Willison's LLM: github.com/simonw/llm
- Simon Willison's files-to-prompt: github.com/simonw/files-to-prompt/tree/0.3
- OpenAI Reasoning Models: platform.openai.com/docs/guides/reasoning/reasoning
- OpenAI Announcement Post: openai.com/index/introducing-openai-o1-preview
- Hacker News Discussion: news.ycombinator.com/item?id=41527143
- Best Prompt Format: youtu.be/W6Z0U11nnhA
- When to Use Prompt Chains: youtu.be/UOcYsrnSNok
- Prompt Chaining Gist: gist.github.com/disler/d51d7e37c3e5f8d277d8e0a71f4a1d2e
🚀 In this video, we dive deep into OpenAI's latest o1 reasoning models, o1-preview and o1-mini, showing you exactly how to leverage them before everyone else! We've been championing prompt chaining and chain of thought techniques, and now OpenAI has embedded these patterns into their new o1 series. Join IndyDevDan as we explore how these reasoning models can revolutionize your AI projects—from AI coding to sentiment analysis on platforms like Hacker News.
🔥 We'll walk through three practical examples demonstrating how to use these new reasoning models effectively. But be warned—the o1 models require a different approach to prompt engineering. We'll use tools from one of our favorite engineers, Simon Willison, to help you get the most out of these models.
📈 See how our AI predictions have come true, and learn how to stay ahead in this rapidly evolving field.
🤖 Whether you're into AI prediction, AI coding, Prompt Engineering or just fascinated by the latest advancements from OpenAI, this video is packed with insights you won't want to miss!
📖 Chapters
00:00 We were right - the prompt chaining based o1 series is here
01:28 o1-preview vs Claude 3.5 - Generating YouTube Chapters
02:35 o1-mini - Setup Simonw's CLI LLM library
07:58 o1-preview - AI Coding Meta Review
16:35 o1-preview - Hacker News Sentiment Analysis
27:44 What's Next - The Future with Reasoning Models
#openai #promptengineering #aicoding
🔗 More AI Coding with AIDER
youtu.be/ag-KxYS8Vuw
🚀 More AI Coding with Cursor
youtu.be/V9_RzjqCXP8
In this video, we showcase the power of the best AI coding assistants, Aider and Cursor. We use them TOGETHER. Whether you're a seasoned engineer or just starting out, this video is packed with insights and techniques to help you ship more code in less time with AI Coding Assistants.
🔧 What You'll Learn:
- AI Coding Assistants: Discover how Aider and Cursor can handle the heavy lifting for you in a new Bun codebase.
- Boost Productivity: Learn the secret sauce of AI coding that can significantly enhance your engineering productivity.
- Configuration Mastery: Get up and running with Aider across codebases using its dot configuration YAML file.
- TypeScript Tips: Explore efficient ways to manage your TypeScript types and improve your codebase structure.
- Notion API Integration: Watch as we set up a Notion API wrapper class to run CRUD operations on Notion pages.
💡 Key Highlights:
- Prompt Caching & Auto Tests: Enable Sonnet 3.5 prompt caching and auto test flags to streamline your coding process.
- Conventions File: Utilize conventions to guide your AI coding assistant in writing consistent and high-quality code.
- Multi-File Prompts: Leverage Aider's ability to update multiple files simultaneously, ensuring your code is always validated and error-free.
- Real-Time Fixes: See how Aider automatically detects and fixes errors, reducing cognitive load and allowing you to focus on the bigger picture.
🔥 Why Watch?
- AI Coding Efficiency: Experience the future of coding with AI, where tools like Aider and Cursor enable you to think less about individual lines of code and more about the overall architecture and design.
- Practical Demonstrations: Follow along as we walk through real coding challenges, showcasing the seamless integration of AI coding assistants into your projects.
- Comprehensive Insights: Gain valuable knowledge on how to use AI tools to their maximum potential, preparing you for the next leap in AI capabilities.
🌟 Stay Ahead of the Curve:
- Like and Subscribe: Don't miss out on more insights and tutorials on AI coding, engineering productivity, and the latest in AI technology.
📖 Chapters
00:00 Action Packed AI Coding Devlog
01:04 Configuring Aider for Optimal AI Coding
02:03 Conventions File: Guiding AI Code Generation
02:30 Setting Up the Project Structure
05:12 First Aider Prompt - setup bun main
06:10 Creating the Notion Wrapper Class
07:15 Auto testing with Aider
9:55 Building a CLI Application with Commander
11:13 AI Coding ADD, Delete notion block
13:50 SECRET SAUCE of AI Coding
15:20 Automatic test resolution with Aider
20:15 Multi-file prompting - Get page blocks function
22:17 AI Coding pattern - Documentation as context
25:25 Improving notion blocks - recursion
29:26 Not Aider vs Cursor - Aider AND Cursor
33:52 Use many GenAI Tools not one - think top three
35:12 Reflections on AI Coding and Future Course Announcement
#aicoding #aiprogramming #coding
🍓 AIDER
https://aider.chat/
🖼️ Mermaid JS AI Agent
github.com/disler/mermaid-js-ai-agent
🔐 Great builder.io article on Cursor, OSS, and Lock-in
builder.io/blog/oss-consequences
💻 Our Cursor Composer Breakdown
youtu.be/V9_RzjqCXP8
⏰ Aider Review 1 YEAR Ago
youtu.be/MPYFPvxfGZs
Unlike Cursor, Aider is open-source and completely free, offering you more control and customization over your AI coding process.
With support from multiple LLM providers and incredible insights from its creator, Paul, Aider is designed to keep you ahead of the AI Coding curve.
🥚🥚🐣 Don't put all your eggs in one basket - explore the benefits of open-source AI coding tools like Aider. Whether you're a seasoned dev or just starting out, this video will show you how to:
- Use Aider's terminal-based interface for precise control over your AI coding process
- Implement the "Ask - Draft - Change" pattern for more accurate code modifications
- Develop new features for our Mermaid JS Diagramming AI Agent with minimal manual input, leveraging Aider's advanced AI coding capabilities
🎥 In This Video:
- Discover why Aider, the original LM-based AI coding tool, often outperforms Cursor.
- Learn how to leverage Aider's multi-file editing capabilities to enhance your coding efficiency.
- See a live demo of building a Mermaid AI Agent to create stunning diagrams effortlessly.
Understand the importance of diversifying your AI coding tools in the rapidly evolving generative AI landscape.
🛠️ Key Topics:
AI Coding: Experience the power of AI in coding with Aider.
Multi-File Editing: Seamlessly edit multiple files with ease.
Diagram AI Agent: Use Mermaid.js + Generative AI to generate and iterate on diagrams quickly.
Human-in-the-Loop: Enhance your agentic coding with iterative feedback and adjustments.
Open Source: Enjoy the freedom and flexibility of an open-source AI Coding tool, AIDER.
🌟 Stay ahead in the world of AI programming and AI software engineering by subscribing to our channel.
Stay focused and keep building.
📖 Chapters:
00:00 Cursor Pop off BUT BEWARE
00:30 Aider - The Original LM-based AI Coding Tool
01:27 Mermaid Diagram AI Agent
04:10 AI Coding with Aider
11:35 Multi-File AI Coding with Aider
17:45 Cheaper LLMs and 100m context window is coming
18:50 Aider gives you incredible AI Coding Insights
#aicoding #programming #aiprogramming
🔗 Reusable OpenAI Fine-Tune Codebase
github.com/disler/reusable-openai-fine-tune
🔗 Black Forest Labs
blackforestlabs.ai
🧠 OpenAI GPT-4o Fine-Tune
openai.com/index/gpt-4o-fine-tuning
🧠 Replicate Black Forest Labs Flux-Pro
replicate.com/black-forest-labs/flux-pro
🚀 Welcome back, engineers! In this video, we dive into the incredible world of fine-tuning with the newly released GPT-4o. Have you ever wondered why you should fine-tune a model and when it makes sense to do so? We’ve got you covered!
🔥 We’ll start with a discussion of the state-of-the-art fine-tuned GPT-4o model by Cosine Genie (ai software engineer) that’s setting new benchmarks on the software engineering verified benchmark. Discover how fine-tuning can bring game-changing performance gains and cost savings, potentially transforming your applications from good to exceptional.
🛠️ Here’s what you can expect:
- Legit Fine-Tuned Use Case: See a real-world application of a fine-tuned GPT-4o model.
- When and Why to Fine-Tune: Understand the key scenarios where fine-tuning can be beneficial.
- Fine-Tuned Code Base: Get hands-on with a code base designed for fine-tuning any OpenAI model, including the latest GPT-4o.
📸 We’ll showcase "Vision Grid", an unreleased tool that uses fine-tuned GPT-4o to create stunning image prompts. Watch as we generate INSANELY, STUPIDLY, high-quality images using Black Forest Labs' Flux 1 image generation models, demonstrating the power of fine-tuned models in action.
💡 We're talking prompt to prompt to images so whether you’re a prompt engineer, AI enthusiast, or software developer, this video is packed with insights on leveraging fine-tuning to enhance your GenAI projects. Learn how to reduce token usage, handle complex domain-specific tasks, and achieve consistently specific outputs.
👍 Hit the like button, subscribe, and join us on this journey as we push the boundaries of generative AI. Stay ahead of the curve with cutting-edge techniques and practical applications that make your work more impactful and efficient.
🛠️ Bonus: Get access to our reusable OpenAI fine-tuning codebase to jumpstart your own projects!
🍓🍓 Fusion Prompt Chain
youtu.be/0Z2BQPuUY50
🍓 Prompt Chaining
youtu.be/UOcYsrnSNok
#promptengineering #midjourney #llm
AI Coding Devlog
youtu.be/1IK69XZZegU
Cursor Copilot++ (Cursor Tab)
youtu.be/Smklr44N8QU
Start AI Coding
cursor.com
Discover how Cursor's groundbreaking Composer feature is revolutionizing AI coding and software development. In this video, we dive deep into the power of multi-file editing and show you how to leverage this game-changing tool to boost your productivity.
🚀 Witness the future of coding as we build a simple prompt editing tool using Cursor's Composer, demonstrating its incredible capabilities across multiple files and components.
🔥 Key highlights:
- Multi-file editing with AI assistance
- Real-time code generation and refactoring
- Seamless integration with existing codebases
- Comparison with other AI coding tools like Aider
- Comparison between the new chatgpt-4o-latest model to claude-3.5-sonnet
🛠️ We'll walk you through:
- Enabling and using the Composer feature
- Creating and modifying Vue.js/Nuxt components
- Implementing styling changes across multiple files
- Resolving errors and debugging with AI assistance
💡 Learn why Cursor's Composer and Aider are leading the pack in AI coding tools, and how they can help you write code faster and more efficiently than ever before.
Whether you're a seasoned developer or just getting started with AI-assisted coding, this video will show you how to harness the power of Cursor's Composer to take your productivity to the next level.
📖 Chapters
00:00 Cursor Composer
00:50 Enable and Setup Cursor Composer
03:05 Multi-file editing
10:05 Move up the stack - focus on what not how
19:55 Big Takeaways for AI Coding Tools
#aicoding #programming #aiprogramming
🔴 Personal AI Assistant Video
youtu.be/ikaKpfUOb0U
🔗 OpenAI AI Agent Codebase:
github.com/disler/personal-ai-starter-pack/tree/structured-outputs
In this video, we talk and showcase how you can leverage structured outputs to build highly reliable, domain-specific AI agents.
🚀 What's New?
OpenAI has introduced structured outputs for function calling and data models, ensuring 100% guaranteed responses in your specified format. Plus, the new GPT-4-O model offers a 50% cost reduction on inputs and a 33% reduction on outputs, making it more affordable to build and run your AI agents.
🔥 Key Highlights:
1. Structured Outputs for Function Calling: Wrap your method parameters using Pydantic or Zod, and get guaranteed responses in your desired format.
2. Structured Outputs for Data Models: Specify the exact object structure you want, and receive 100% guaranteed responses.
3. Cost Reduction: The state-of-the-art GPT-4-O model cuts input costs by 50% and output costs by 33%, making it easier and cheaper to build domain-specific agents.
🛠️ Demo Time!
Watch as we demonstrate an OpenAI exclusive super agent in action. Using speech-to-text (stt) and text-to-speech (tts) capabilities, our personal AI assistant, Ada, performs tasks like generating images, converting formats, resizing, and opening directories—all through natural language commands. See how easy it is to interact with your digital assistant and get work done faster than ever.
🌟 Why You Should Care:
Building AI agents that are domain-specific and highly reliable has never been simpler. Whether you're a software engineer, a tech enthusiast, or an AI pioneer, understanding and utilizing these new features can position you at the forefront of innovation.
💡 Get Started:
- Like and Subscribe: Stay ahead of the curve with more insights on AI technology.
- Code Base: Check out the code base link in the description to get started quickly.
- Example Code: We've included examples from OpenAI's blog to help you implement these features effortlessly.
Join us as we explore the future of AI with structured outputs, making your personal AI assistant more powerful and efficient. Let's innovate together and transform your approach to AI with every video.
Stay focused and keep building.
📖 Chapters
00:00 OpenAI is the monopoly for AI Agents
01:38 AI Agent Demo with Structured Outputs
04:05 Breakdown OpenAI Personal AI Assistant
07:07 Is Strawberry a GPT Next AI Agent?
09:25 Reflect, Improve, Act - What's after AI Agents?
#aiagents #aicoding #aiassistant
🔑 Unlock the power of voice-assisted work with your very own personal AI assistant! In this video, we break down the key components and show you how to build a lightning-fast, customizable AI assistant using cutting-edge speech-to-text and text-to-speech technology.
📊 See real performance benchmarks and find out which setup delivers the fastest results for transcription, thinking, and voice generation.
🔥 We compare three powerful combinations:
1. Groq + ElevenLabs
2. Pure OpenAI implementation
3. AssemblyAI + ElevenLabs
Let's focus in on how to build your own personal AI assistant:
🛠️ Key topics covered:
- Speech-to-text (STT) and text-to-speech (TTS) technologies
- Leveraging GPT-4O Mini for rapid responses
- Voice options with ElevenLabs
- Building a reusable Personal AI Assistant class
- Crafting the perfect AI assistant prompt
🧠 Learn how to escape the walled gardens of big tech and take control of your own AI productivity tools. Perfect for software engineers looking to supercharge their workflow!
⚡ Ready to revolutionize your productivity? Hit subscribe and join us as we push the boundaries of what's possible with personal AI assistants. Don't miss out on this game-changing technology that's transforming the way we work!
💻 Personal AI Assistant STARTER PACK
github.com/disler/personal-ai-starter-pack
📖 Learn the best prompt format
youtu.be/W6Z0U11nnhA
🔗 Resources
Groq STT: console.groq.com/docs/speech-text
ElevenLabs TTS: elevenlabs.io
OpenAI STT: platform.openai.com/docs/guides/speech-to-text
OpenAI TTS: platform.openai.com/docs/guides/text-to-speech
AssemblyAI STT: assemblyai.com
🍞 Chapters
00:00 Groq STT, ElevenLabs TTS, Cringe Assistant
03:15 Personal AI Assistants
04:48 Three Key Components of AI Assistants
05:23 Pure OpenAI AI Assistant
08:28 Breakdown STT, think, TTS, Assistant Prompt
12:30 AssemblyAI STT, ElevenLabs TTS Slow Assistant
14:57 Results - Speed Comparison of AI Assistants
18:00 High Level Code Breakdown
🚀 Ready to unlock the true potential of your AI agents? In this video, we're diving deep into the world of prompt formats to find out which one reigns supreme: Markdown, XML, or Raw Prompts. Whether you're a seasoned AI engineer or just starting out, understanding the best prompt format can drastically improve your AI workflows and performance.
🔥 In the past few weeks, we've seen incredible advancements in large language models like Llama-3.1 8B, Llama-3.1 405B, and Mistral Nemo. These models are pushing the boundaries of AI capabilities, and we're here to explore how different prompt formats can optimize their performance. Inspired by Anthropic's XML format insights, this video will help you master prompt engineering and choose the best format for your AI agents.
🛠️ We're not just talking theory; we're running real tests with the Prompt Testing Framework Promptfoo! Watch as we compare Markdown, XML, and Raw Prompts across multiple models, including the latest Llama-3.1 405B and Mistral Nemo. We'll show you how each format performs in various scenarios, from simple bullet summaries to complex YouTube chapter summaries and AI command selectors.
🌟 Hit the like and subscribe for more insights on prompt engineering, AI agents, and agentic workflows. Stay ahead of the curve and transform your AI projects with the best prompt formats.
💡 Key takeaways:
- Prompt Format: Discover why XML tags might be the best format for maximizing prompt performance and accuracy.
- Surprising Results: See how well raw prompts perform against structured formats (hint: it's not as bad as you think)
- Markdown Prompt: Learn when to use Markdown for readability and ease of use.
- Raw Prompts: See how raw, unformatted prompts stack up against structured formats.
- Llama-3.1: Get insights into the latest models like Llama-3.1 8B and 405B, and how they handle different prompt formats.
- Mistral Nemo: Explore the performance of Mistral Nemo in various prompt scenarios.
- AI Agent: Understand how to build robust AI agents with the right prompt formats.
- Agentic Workflow: Learn how to optimize your agentic workflows for better results.
Join us as we put these prompt formats to the test and find out which one is the best for your AI agents. Whether you're working on AI coding assistants, personal AI assistants, or complex agentic workflows, this video is packed with actionable insights to help you succeed.
🔗 Resources
📄 Why use XML tags docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/use-xml-tags#why-use-xml-tags
🧪 Start testing your prompts github.com/disler/elm-itv-benchmark
🎥 Previous promptfoo testing video docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/use-xml-tags#why-use-xml-tags
🎓 GPT-4o mini SOTA Accuracy TRICK youtube.com/watch?v=0Z2BQPuUY50
📢 Mistral AI large enough mistral.ai/news/mistral-large-2407
📖 Chapters
00:00 Wow Llama 3.1, Mistral Large 2, and Gpt-4o mini
01:25 What's the best prompt format?
02:48 Promptfoo Test 1 - Bullet Summary
05:00 Promptfoo Test 2 - YouTube chapter summaries
07:18 Promptfoo Test 3 - Personal AI Commands
10:03 Promptfoo Test 4 - Nuxt Vue Component Generation
11:11 Promptfoo Test 5 - Code Debugging
14:20 Promptfoo Test 6 - Update Config File
15:50 Promptfoo Test 7 - QA Chatbot
17:03 Promptfoo Test 8 - Script to Key Ideas
18:30 Why use XML? Why use Markdown?
#prompt #agentic #testing
Are you ready to revolutionize your AI workflows without breaking the bank? Introducing GPT-4O Mini, the game-changing model that's making high-performance AI accessible to everyone!
In this video, we dive deep into the staggering performance of GPT-4O Mini, a cost-effective model that rivals state-of-the-art giants like GPT-4 and Claude 3.5 Sonnet. Discover how this affordable intelligence solution can deliver impressive results at a fraction of the cost. 🚀
🚀 Learn how to leverage GPT-4O Mini with advanced techniques like prompt chains and fusion chains to achieve state-of-the-art results at a fraction of the cost.
🔥 Key topics covered:
- GPT-4O Mini vs. GPT-4 and Claude 3.5 Sonnet performance comparison
- Prompt chaining techniques for enhanced results
- Fusion chains: combining multiple model outputs for optimal performance
- Real-world application: Building an intelligent content recommendation system
👨💻 Watch as we build "Zero Noise," an agentic application that filters and recommends relevant content using GPT-4O Mini prompt chains and fusion chains. See how this affordable model powers complex workflows, including:
- Keyword extraction from scraped content
- Intelligent filtering based on user feedback
- SEO-driven content recommendations
🔧 Dive into the code and see how to implement these techniques in your own projects. Learn how to create living software that works while you sleep!
If you're interested in prompt engineering, AI agents, and building intelligent software that leverages the latest in LLM technology, this video is a must-watch. Hit like and subscribe to join us on the journey of creating intelligence that works on our behalf!
🔗 Resources
💻 Minimalist Prompt Chain + Fusion Chain Code: gist.github.com/disler/d51d7e37c3e5f8d277d8e0a71f4a1d2e
🙏 Fusion Chain Video: youtu.be/iww1O8WngUU
🤔 When to use Prompt Chains: youtu.be/UOcYsrnSNok
#agentic #promptengineering #aiengineer
🚀 Yeah so prompt chaining is legit. Let's push it EVEN FURTHER BEYOND and discover the power of fusion chains and multi-chain techniques to maximize the potential of Large Language Models (LLMs) to get next generation LLM performance (GPT-5, Claude 4, and Gemini 2).
In this video, we explore how prompt chains can transform your approach to AI, allowing you to chain together multiple prompts to enhance reasoning and decision-making. We'll break down the concept of the prompt chain, where the output of one prompt becomes the input of the next, creating a powerful sequence that mimics human workflows. This is nothing new, we've covered it on the channel and if you use GenerativeAI you already utilize prompt chains.
What's next is what happens when you multiply the number of chains you have, evaluate and FUSE the outputs to get the best possible result. This takes 'lets think step by step' and multi-agent reasoning to a new level.
🔥 Fusion Chains take your prompt chains a step further by running multiple chains simultaneously and merging their outputs to achieve the best possible result. Imagine having multiple AI agents from OpenAI, Anthropic, and Google working together to provide you with the most accurate and efficient outcomes. This technique, also known as beam chaining or the competition chain, is a game-changer in the world of AI.
🛠️ Watch as we demonstrate practical applications of these techniques, showcasing how to build agentic workflows that operate seamlessly and efficiently. From agentic software to multi-chain outputs, we'll show you how to leverage these advanced patterns to create powerful AI-driven tools and applications.
🌟 Join us as we tackle key questions in the AI community:
- Does adding multiple chains improve performance?
- Will future models like GPT-5, Claude 4, and Gemini 2 make prompt chains obsolete?
- What is the optimal flow for building agentic workflows?
💡 Whether you're a software developer, AI enthusiast, or indydevdan follower (lets goooo), understanding these concepts will position you at the forefront of AI innovation. Discover how prompt chains and fusion chains can elevate your GenAI projects, making your work more impactful and future-proof.
Hit the like and subscribe for more insights on how to master prompt chaining and agentic workflows. Stay ahead of the curve and transform your approach to AI with every video. We're on the golden path to building LIVING SOFTWARE.
Subscribe now and join us on this journey to mastering the future of AI!
✅ Let's build Agentic Workflows
Part 4: youtu.be/1IK69XZZegU
Part 3: youtu.be/F0eOYrA6ggY
Part 2: youtu.be/UOcYsrnSNok
🔗 Resources
📝 Minimalist Prompt Chain & FusionChain gist gist.github.com/disler/d51d7e37c3e5f8d277d8e0a71f4a1d2e
🧠 Big AGI Beam big-agi.com/blog/beam-multi-model-ai-reasoning
📚 "More Agents Is All You Need" research paper arxiv.org/pdf/2402.05120
📖 Chapters
00:00 The Prompt
00:34 The Prompt Chain
02:48 The Fusion Chain
04:50 Prompt Chaining Questions
11:05 Minimalist Prompt Chain API
12:02 Fusion Chain API
12:50 Zero Noise LEARN Agentic Workflow
#agentic #promptengineering #aiagents
In this AI coding devlog, we're building "Zero Noise," an agentic workflow that filters content for you using the power of AI!
🚀 We'll be using Aider AI with the powerful Sonnet 3.5 model to code a tool that scrapes your favorite websites, blogs and changelogs and alerts QUITELY Notifies you to new and relevant updates.
🔥 Learn how to:
- Leverage AI coding assistants for faster, more efficient software engineering.
- Implement agentic workflows to automate tasks and boost productivity.
- Craft effective prompts and prompt chains to get the most out of Sonnet 3.5.
- Build a personalized content curation system to filter out the noise.
💡 This video covers key concepts like:
- Live showing of AI coding best practices and techniques.
- Aider AI setup and usage for AI-powered development.
- Agentic engineering structures for building intelligent systems.
- Prompt engineering in markdown format for optimal AI interaction and readability.
Don't waste time sifting through irrelevant information.
Like - Sub - Join the journey as we become Agentic Engineers.
🔗 Resources
- Part 3 - Auto Updating Blog - youtu.be/F0eOYrA6ggY
- Part 2 - Prompt Chains - youtu.be/UOcYsrnSNok
- Part 1 - Master the prompt - youtu.be/4hSFcjspGOw
- Start Coding With AI: https://aider.chat
📖 Chapters
00:00 AI Coding a Zero Noise Info Curation Tool
01:19 Reviewing scraping and streamlit
02:05 Using Aider and Sonnet 3.5
04:01 Importance of Curating Information
07:37 Prompting Agentics Prompt Chains (lol)
10:30 Markdown Prompt Chains
13:35 Aider generates Pydantic Models from JSON
19:17 Detecting Changes in Content
20:27 SUCCESS - Updates detected
21:00 Act Step in the Agentic Workflow
22:34 Running the Workflow
23:35 Adding Aider blog json with Aider
Discover how to create powerful agentic workflows that can run, prompt, and report automatically by looking at a real use case. This video explores a piece of the future of engineering, showing you how to build systems that work for you while you sleep.
🔬 We dive deep into:
• The structure of sequential agentic workflows
• How to use prompt chains for complex reasoning
• Techniques for auto-updating content with AI
• Integrating notifications into your agentic workflows
🚀 Learn to harness the power of:
• Build agentic workflows that can make decisions and take action
• Claude 3.5 Sonnet insane accuracy and instruction following
• Prompt engineering for real-world use cases
👨💻 Perfect for:
• Engineers obsessed with automation
• AI engineers and prompt engineers
• Developers exploring agentic systems
🔑 Key takeaways:
• Step-by-step breakdown of agentic workflow components
• Tips for building robust, self-improving agentic workflows
• Steps to create "living software" that evolves over time
Don't miss this in-depth look at the future of AI development. Like, subscribe, and hit the notification bell to stay updated on the latest in agentic engineering and AI workflows!
🔥 When to use prompt chains:
youtu.be/UOcYsrnSNok
✅ Minimalist Prompt Chain Code
gist.github.com/disler/d51d7e37c3e5f8d277d8e0a71f4a1d2e
🔗 Resources
Pub/Sub Notifications: https://ntfy.sh/
Python Schedule: schedule.readthedocs.io/en/stable
📖 Chapters
00:00 Prompts, Prompt Chains, Agentic Workflows
01:23 Auto Blog Agentic Workflow Demo
02:10 NTFY - Crucial Notification Tool
03:20 Breaking Down the Workflow Steps
05:43 Agentic Step - Running the Prompt Chain
10:30 Sequential Agentic Workflow Steps
14:59 5 Tips for Building Agentic Workflows
16:54 The Future - Less Code, More Agent Work
17:40 LLM Killer User Case - Living Software
#agentic #promptengineer #aiagents
Are you curious about when to use prompt chains and why startups are moving away from Langchain and other LLM libraries? This video dives deep into these topics and reveals the minimalist prompt chaining method that can revolutionize your productivity.
🚀 In this video, we're breaking down the ULTIMATE guide to prompt chains using Claude 3.5 Sonnet, Anthropic's latest powerhouse LLM. Learn why startups are ditching complex libraries like Langchain and Autogen in favor of raw, unfiltered prompts.
🔥 Unlock the potential of minimalist prompt chaining and see how it can skyrocket your productivity. We'll show you:
1. A step-by-step breakdown of our minimalist chainable API
2. 4 crucial questions to determine when you should use prompt chains
3. The pitfalls of over-relying on LLM libraries and frameworks
💡 Discover why staying close to the metal with your prompts is CRITICAL in the ever-evolving AI landscape. We'll demonstrate how to build valuable prompt chains without unnecessary abstractions, giving you full control over your AI agents.
⚡️ Watch as we transform a simple factorial calculator into a powerful teaching tool using our minimalist approach. Plus, get a sneak peek at a production-level prompt chain driving a full agentic workflow!
🔧 Whether you're building AI coding assistants, research tools, or personal AI helpers, mastering prompt chains is your ticket to creating next-level agentic applications. Don't get left behind in the AI revolution!
🎓 Ready to level up your prompt engineering skills? Hit subscribe and join us on this journey to becoming an agentic engineering pro. Let's harness the true power of Claude 3.5 Sonnet and build AI agents that work tirelessly for you and your users.
💼 Remember, in the world of AI, the prompt is king. Don't give away your most valuable asset to complex libraries. Stay agile, stay close to the metal, and unlock the full potential of your AI workflows with prompt chains!
Like, subscribe, and comment with your thoughts on prompt chains and agentic workflows.
Let's COOK.
💻 Minimalist Prompt Chain Code
gist.github.com/disler/d51d7e37c3e5f8d277d8e0a71f4a1d2e
🔴 Master the prompt (Top 5 Elements)
youtu.be/4hSFcjspGOw
🔗 Resources:
- Octomind: https://www.octomind.dev/blog/why-we-no-longer-use-langchain-for-building-our-ai-agents
- Langchain: langchain.com
- Simonw Lightweight LLM Library: github.com/simonw/llm
📖 Chapters
00:00 From Prompts to Prompt Chains
01:23 Minimalist Chainable API for Prompt Chains
04:00 Key Benefits of Using Prompt Chains
07:29 Four Guiding Questions for Using Prompt Chains
12:55 Problems with LLM Libraries like Langchain
15:52 Octomind blog post - Libraries are OVERKILL
18:20 Why the Prompt is All That Matters in Generative AI
19:22 Building a Production-Level Agentic Workflow
21:55 Closing Thoughts: Embrace Minimalism in AI Development
#anthropic #langchain #promptengineer
Ever wondered which parts of your prompts ACTUALLY MATTER? Let's simplify the sea of prompt engineering tips and tricks with this concise guide on the five essential elements that gives you 80% of the results with 20% of the effort. I've crafted thousands of prompts, and today, I'm sharing the key elements that consistently deliver top-notch outcomes with minimal effort.
🚀 Discover the five crucial components: model, purpose, variables, examples, and output. These elements form the backbone of effective prompt engineering, helping you achieve 80% of the results with just 20% of the effort.
🔗 These elements enable composability between prompts, allowing you to chain together outputs and inputs to build powerful prompt chains. By focusing on the underlying technology and understanding the prompt at its core, you'll be well-equipped to create AI Agents and agentic workflows.
🔥 In this video, we break down each element with clear examples and practical tips:
Model: Learn why the model you choose has the most significant impact on your prompt's performance.
Purpose: Understand how a clear goal enhances your prompt's effectiveness.
Variables: Master the use of dynamic and static variables to make your prompts adaptable and reusable.
Examples: See how concrete examples can guide your AI to produce the exact output you need.
Output: Explore the importance of structured outputs, like JSON, for building reliable and consistent AI workflows.
🛠️ We'll showcase practical applications of these elements through detailed examples, including a Nuxt.js component and a comprehensive Omnicomplete prompt. Whether you're a developer, AI enthusiast, or product builder, this video is packed with actionable insights to enhance your prompt engineering skills.
🌟 Hit the like and subscribe for more tips on AI agents, agentic workflows, and prompt chains. Stay ahead of the curve and stay plugged into the latest models like GPT-4o, Gemini 1.5 Pro, and other state-of-the-art models to use their full potential for your prompts, AI Agents, and agentic workflows.
Keep prompting, keep building and stay ahead of the curve.
🔗 Resources
Llama-3 70b Omnicomplete: youtu.be/28zuliyLd5Q
LLM OS: youtu.be/8wSH4XukcH8
7 Prompt Chains: youtu.be/QV6kaNFyoyQ
📖 Chapters
00:00 Maximizing Prompt Value
00:34 The Five Key Elements of Prompts
01:24 E1 - Model - Why the Model Matters Most
01:59 E2 - Purpose - Defining Your Goal
02:25 E3 - Variables - Dynamic and Static
03:32 E4 - Examples - Concrete Examples for Clarity
05:49 E5 - Output - Structured and Reliable JSON
07:16 Concise, valuable, reusable prompts
08:00 Building Prompt Chains and AI Agents
08:56 Recap - The 80-20 of Prompt Engineering
09:48 Real Example Prompt - OmniComplete - Static Variables
12:20 Real Example Prompt - Nuxt / Vue Component
13:53 Top 5 Elements for Reusable Prompts
14:30 Focus on the groundwork - the prompt
#promptengineering #aiagents #aiengineering
🌟 In this video, we break down the three essential categories of AI tools that every AI engineer needs to know: AI Copilots, AI coding assistants, and AI software engineers. Learn the differences, capabilities, and why these tools are revolutionizing the way we code.
🛠️ We discuss the power of 'tier 1' AI Copilots like GitHub Copilot and Supermaven, and see how they boost productivity with intelligent autocomplete features. Move up to 'tier 2' AI coding assistants like Cursor, Aider, and Continue, which take coding assistance to a whole new level with advanced capabilities. Finally, we briefly discuss the future with 'tier 3' AI software engineers like Devon and Copilot Workspace, and understand their potential to transform the coding landscape.
🔥 Watch as we demonstrate practical examples and high-level features of these tools, from basic autocompletes to advanced coding prompts. See how tools like Cursor's Copilot++ can streamline your workflow, making you a more efficient and effective coder.
💡 Whether you're a seasoned AI engineer or just starting with AI coding, this video is packed with insights and actionable tips to help you stay ahead of the curve. Don't miss out on the future of coding—embrace these tools and supercharge your productivity!
👍 If you found this video helpful, make sure to hit the like button and subscribe for more content on AI coding, AI engineering, and the latest advancements in generative AI tools. Stay tuned for more exciting updates and tutorials to elevate your coding.
🔥 Previous AI CODING video
youtu.be/YALpX8oOn78
🔗 Resources
- github copilot github.com/features/copilot
- aider https://aider.chat/
- cursor https://cursor.sh/
- bun https://bun.sh/
- opendevin github.com/OpenDevin/OpenDevin
- continue https://www.continue.dev/
📖 Chapters
00:00 AI Copilots, Coding Assistants, and Software Engineers
00:35 Understanding AI Tool Capabilities
01:15 AI Coding Tools
03:35 Key Features of AI Copilots
05:00 Key Features of AI Coding Assistants
12:08 AI Software Engineers are NOT there yet
13:00 Recommendations for AI Coding Tools
#aicoding #agentic #llm
Stop wasting time writing boilerplate code! The future of coding is here, and it's called AI coding assistants.
Maybe you've heard of github copilot, devin, copilot workspace, cursor, or aider. Let's focus on the HIGHEST PRODUCTIVITY AI Coding Tools that you can use AS SOON as you finish this video.
In this video, we're diving deep into two of the BEST AI coding assistants available today: Aider and Cursor. We'll explore their unique features, strengths, and how they can supercharge your coding productivity by up to 2x-10x (and beyond honestly)
Here's what you'll discover:
- Aider in the browser: Experience the power of Aider's new browser-based editor for seamless code generation.
- Aider's multi-file editing: Experience the power of seamless code generation and refactoring across your entire codebase with Aider's groundbreaking multi-file editing capabilities.
- Real-world coding examples: See Aider and Cursor in action as we build a complete application from scratch using clear, concise prompts.
- Unlocking 8x productivity: Learn how AI coding assistants free you from tedious tasks, allowing you to focus on the bigger picture and build better software faster.
📊 Stay ahead of the curve with insights into the future of AI coding. We'll look at the latest blogs from Cursor and Aider, highlighting upcoming features like next action prediction and automatic bug detection.
❌ Don't miss out on the AI coding train. Equip yourself with the best coding AI tools and elevate your engineering game. Hit the like and subscribe buttons to stay updated with more AI coding content.
Stay ahead of the curve with the latest in AI coding, AI code writers, and artificial intelligence coding.
Bro, it's time to pop off w/ai.
💻 AI Copilots vs AI Coding Assistants vs AI Engineers
youtu.be/2j_fgMPJGM0
🔗 Links:
- aider https://aider.chat/
- aider blog https://aider.chat/blog/
- cursor https://cursor.sh/
- cursor blog https://cursor.sh/blog/problems-2024
- bun https://bun.sh/
📖 Chapters
00:00 AI Coding Assistants: Why You Need Them
00:09 Generating Code with One Prompt
00:40 Running the AI Generated Code
02:43 8x Productivity Boost with AI Coding Assistants
04:19 Multi-File Editing with Aider in the Browser
06:48 Don't miss the AI coding train - stay ahead as an engineer
07:32 AI Coding 3 files at once with a CRUD Test
11:20 The Future of AI Coding
12:00 Copilot++ next action prediction with Cursor
15:50 Aider vs. Cursor: Two Approaches to AI Coding
18:30 Our endgame: full agentic tools
#aicoding #typescript #agentic
Are you ready to maximize GenAI in a world of infinite information? 🌐 As an engineer or product builder, it’s crucial to prioritize and utilize LLM Tech effectively for your career and product development. In this video, we tackle essential questions like how to best use LLM Tech, where to focus your time, and how to maximize your potential. 📈
We’ll dive into a simple framework that breaks down the pieces of modern information—text, code, images, and videos—showing you how to prioritize and manipulate these components using prompts and AI agents. 🚀
This is how we'll become an agentic engineer, mastering prompt engineering, prompt chains, and efficient AI workflows. We’ll explore practical examples for different use cases, whether you’re a marketing specialist or a software engineer. Understand how to leverage tools like GPT-4o (GPT-4 Omni) to enhance your work, boost productivity, and achieve high-quality outputs. 🔧💡
This video is a crucial steps to knowing how to use LLM technology effectively for YOUR specific role and use cases.
We'll offer you a framework to help you focus on on the new microsoft, openai, google, and opensource GenAI tools and tech that matter most for you.
On the channel we're obsessed with...
- Agentic technology and workflows
- Best practices with LLMs
- AI-driven development and AI-generated content
- AI for text, image, and video creation
- Gen AI prioritization and strategies
- AI tool utilization and increasing productivity
- AI Coding Assistance, and AI Engineers
- Personal AI Assistants and digital companions
Stay ahead in the rapidly evolving landscape of AI with our GenAI Prioritization framework to improving your economic output and learning ability.
🔔 Subscribe now and Don’t miss out on the future of AI-driven development!
🔥 POP OFF WITH AI.
🔗 Resources For YOU
Prepare for "GPT Next": youtu.be/JBgUmTUQx0I
No, ChatGPT Sky is NOT a resource. It's a...: youtu.be/gdrgFCldvrA
Learn anything with LLMs fast (Fishermans prompt): youtu.be/gS5u3j1OXjg
📖 Chapters
00:00 Introduction to Gen AI and Information Overload
00:14 Key Questions for Engineers and Product Builders
00:55 Framework for Prioritizing GenerativeAI
01:18 The 4 Key Pillars of Modern Information for LLMs
01:50 The Power of Prompts in LLM Technology
03:13 Example Use Case: Marketing Specialist
04:07 Example Use Case: Software Engineer
05:10 The Guiding Question
07:18 MAXing your Learning Ability and Economic Output
08:17 MAXing your Output Quantity and QUALITY
09:40 Empty Youtube Channels, Deep Fakes and Information Filtering
11:02 Using the Generative AI Prioritization Framework
11:34 IndyDevDan Channel Mission - Agentic Engineering
#genai #gpt5 #promptengineering
GenerativeAI is going to revolutionize the way you work and interact with technology. OpenAI just broke the internet with GPT-4o, inside of a beautiful ui/ux, and VOICE with SKY.
In this video, we dive deep into the game-changing release of ChatGPT Sky built on top of GPT-4o, and the world of digital companions and AI assistants, focusing on the groundbreaking advancements with Sky, built on the incredible GPT-4O model by OpenAI.
🌟 Why ChatGPT Sky? Who cares bro?
Sky isn't just an AI assistant; it's a DIGITAL COMPANION that understands, remembers, and connects with you. This video explores how Sky can support your career, streamline your workflow, and provide real-time, multimodal interaction like never before. We’ll discuss the profound differences between digital companions and traditional AI assistants, highlighting why Sky with GPT-4o stands out.
🔍 Key Topics Covered:
- Sky and Digital Companionship: Understand the emotional and functional benefits of having a digital companion versus a personal AI assistant.
- The Future of Generative AI: Explore the latest trends from OpenAI and Google, including GPT-4O, Gemini Pro, and their implications for the future.
- Capitalizing on AI Technology: Learn strategies to leverage these advancements for personal and professional growth, with a focus on prompt engineering and AI agents.
🔥 Highlights:
- OpenAI’s GPT-4o: Discover why this model is a potential precursor to GPT-5 and how it enhances Sky’s capabilities.
- Multimodal Interaction: See how Sky utilizes multimodal functionality for seamless, real-time responses.
- AI Assistant vs. Digital Companion: Learn the critical differences and why digital companionship is the future.
🛠️ Practical Strategies:
- Implementing prompt engineering techniques to maximize AI efficiency.
- Building and using AI agents to automate tasks and enhance productivity.
- Understanding the impact of multi-modal models and large context windows on AI performance.
💡 Stay Ahead of the Curve:
Don’t miss out on the opportunity to stay at the forefront of AI technology. Whether you’re a software developer, a tech enthusiast, or simply curious about the future of AI, this video is packed with insights and practical tips to help you harness the power of digital companions and generative AI.
👉 Subscribe for more cutting-edge content on AI, digital companions, and prompt engineering. Hit the like button if you find this video valuable.
Keep building, and stay focused.
🔗 Catch Up & Keep Up
- Prepare for 100x: youtu.be/JBgUmTUQx0I
- OpenAI GPT-4o Release: openai.com/index/hello-gpt-4o
- Build your own AI Assistant (ADA): youtu.be/kLi4SKlc4HQ
📖 Chapters
00:00 Digital Companionship: The Future of Work
01:24 OpenAI is Hiding Something - GPT-4 Omni is a Soft Launch of GPT5
02:19 Sky - The Rise of the Digital Companion
03:00 Big Ideas - Sky, Future of GenAI, Capitalization
03:30 Digital Companions vs AI Assistants
06:20 The Future of Generative AI - Trends and Predictions
08:15 GPT-4 Omni, Gemini, and Project Astro - What's Next?
09:57 Capitalizing on Generative AI - Strategies for Success
12:17 Building a Work-Oriented Relationship with Your Digital Companion
14:20 Your Data is Your Most Valuable Asset - The Future of UX
15:46 Fishy Benchmarks - Hitting the GPT Limit? Or GPT5 Preview?
#gpt5 #promptengineer #aiassistant
Writing autocomplete code is a challenge. Then you have to write it again and again as the business logic changes. 🔥 In this video, I'll show you how to harness the power of LLM Llama 3-70b with Groq to create an OmniComplete – a self-improving, domain-agnostic autocomplete that works across ALL your tools and applications! 🤯
Imagine this: Your users start typing, and your OmniComplete instantly suggests relevant completions based on ALL previous user inputs AND your unique domain knowledge. 🤯 No more rigid dropdowns or limited suggestions – this is next-level LLM autocomplete!
Here's what you'll discover:
- The HUGE difference between traditional autocompletes and LLM-powered autocompletes – and why LLMs are GAME-CHANGING!
- How LLM AutoCompletes self-improve with every use – watch your autocomplete get smarter over time!
- Actionable insights from your users – uncover what your audience REALLY cares about, directly from their autocomplete interactions!
- A simple yet POWERFUL prompt-centered architecture – easily reuse your OmniComplete across different domains with minor prompt tweaks!
The complete codebase – get up and running with your OWN OmniComplete today!
Plus, we'll dive deep into:
- Prompt engineering for autocompletion – craft prompts that deliver spot-on suggestions.
- One-shot prompts – get accurate completions with just a single example.
- Building a prompt-centered architecture – design a system that revolves around your prompts for maximum flexibility and reusability.
- Prompt testing and validation – ensure your OmniComplete is always delivering high-quality results.
Ready to supercharge your user experience with AI-powered autocomplete? 🚀 Hit that like button, subscribe, and let's build the future of autocomplete together!
---
What do you predict OpenAI will release today (May 12th, 2024) ? My Prediction ⬇️⬇️⬇️
Prediction #1: OpenAI will announce an on device compatible, GPT4 level model.
Prediction #2: OpenAI will announce apple as a partner and discuss plans to deploy ‘GPT4-mini’ on the iPhone.
---
🔗 Resources
Codebase: github.com/disler/omni-complete
Learn about BAPs (Big Ass Prompts): youtu.be/JBgUmTUQx0I
Master Prompt Testing: youtu.be/sb9wSWeOPI4
Build better prompts: youtube.com/watch?v=wDxZhkQj27Y
Unocsss: https://unocss.dev/
📖 Chapters
00:00 Increase your earnings potential
00:38 Omnicomplete - the autocomplete for everything
01:16 LLM Autocompletes can self improve
02:00 Reveal Actionable Information from your users
03:20 Client - Server - Prompt Architecture
05:30 LLM Autocomplete DEMO
06:45 Autocomplete PROMPT
08:45 Auto Improve LLM / Self Improve LLM
10:25 Break down codebase
12:28 Direct prompt testing integration
14:10 Domain Knowledge Example
16:00 Interesting Use Case For LLMs in 2024, 2025
#promptengineering #llama3 #autocomplete
If you don't have a plan, how will you know you're succeeding? In this video, we create a plan and discuss tactics for preparing for a 100x improvement in SOTA LLMs, state of the art large language models.
🚀 Let's discuss the Large Language Models (LLMs) and the mind-blowing potential of a 100x improvement in AI capabilities. Inspired by insights from Sam Altman, this video unpacks the game-changing advancements in models like GPT-4, Claude Opus, and Gemini Pro. Discover how the future might look with GPT-5 or even beyond!
🔥 In today's tech-driven world, prompt engineering, prompt chains, and context-filled prompts are not just jargon—they are the backbone of high-performance AI workflows. Whether you're a software developer, a tech enthusiast, or an AI pioneer, understanding the power of 'Big Ass Prompts' (BAPs) and the significance of a large context window can position you at the forefront of innovation.
🛠️ We're not just theorizing; we're applying! Watch as we demonstrate practical strategies to leverage your current tools to their maximum potential, preparing you for the upcoming 100x leap in efficiency and effectiveness. From expanding your problem set with ingenious prompt chains to mastering one-shot and few-shot prompts, this video equips you with the tools to excel in the age of advanced AI.
🌟 Hit the like and subscribe for more insights on how you can transform your engagement with AI technology, making your work in programming, knowledge work, and beyond more impactful than ever. Stay ahead of the curve, prepare for the future with us, and turn these insights into actions that catapult your skills and solutions into a new era of AI prowess.
💡 Remember, the key to mastering the future of AI is not just about understanding the technology but being ready to implement and scale it effectively. Join us as we pave the way to a smarter, more efficient world powered by next-generation LLMs like GPT-5 and beyond.
Subscribe now and transform your approach to AI with every video—let's innovate together!
🎥 Featured Media
20VC Sam Altman & Brad Lightcap
youtu.be/G8T1O81W96Y?si=lrN5eELxq9-Z0gaT&t=1249
📖 Chapters
00:00 The 100x LLM is coming
01:30 A 100x on opus and gpt4 is insane
01:57 Sam Altman's winning startup strategy
03:16 BAPs, Expand your problem set, 100 P/D
03:35 BAPs
06:35 Expand your problem set
08:45 The prompt is the new fundamental unit of programming
10:40 100 P/D
14:00 Recap 3 ways to prepare for 100x SOTA LLM
#promptengineer #gpt5 #ai
How do know when it is?
Using the ITV Benchmark with Llama 3, Gemma, PHI 3, you can be 100% sure that the ELM is ready for your use case.
Let's make 1 thing absolutely clear: The cost of the prompt is going to ZERO.
The world of AI is evolving at a BREAKNECK pace, and the latest advancements in efficient language models (ELMs) like Llama 3, Gemma, OpenELM, and PHI 3 are pushing the boundaries of what's possible with on-device AI. 🤖💡
LLama 3 8b, and Llama 3 70b have hit the top 20 on the LMSYS Chatbot Arena Leaderboard in less than a week of launch. You can bet that the open source LLM community is tweaking and tuning llama3 to make it even better. It's likely we'll see the 8k context window improved to 32k and above in a matter of days.
But with so many options and rapid developments, how do you know if an ELM (efficient language model aka on device language model) is truly ready for YOUR specific use case? 🤔
Enter this video and the ITV Benchmark - a powerful tool that helps you quickly assess the viability of an ELM for your needs. 📊💪
In this video, we dive deep into the world of ELMs, exploring:
✅ The key attributes you should consider when evaluating an ELM, including accuracy, speed, memory consumption, and context window
✅ How to set your personal standards for each metric to ensure the ELM meets your requirements
✅ A detailed breakdown of the ITV Benchmark and how it can help you determine if an ELM (llama3, phi3, gemma, etc) is ready for prime time
✅ Real-world examples of running the ITV Benchmark on Llama 3 and Gemma to see how they stack up 🥊
✅ Gain access to a hyper modern, minimalist prompt testing framework built on top of Bun, Promptfoo, and Ollama
We'll also discuss the game-changing implications of ELMs for your agentic tools and products. Imagine running prompts directly on your device, reducing the cost of building to ZERO! 💸
By the end of this video, you'll have a clear understanding of how to evaluate ELMs for your specific use case and be well-equipped to take advantage of these incredible advancements for both LLMs and ELMs. 🚀
ELMs, setting standards and clean prompt testing enable you to stay ahead of the curve and unlock the full potential of on-device AI! 🔓💡
Like and subscribe for more cutting-edge insights into the world of AI, and let's continue pushing the boundaries of what's possible together! 👍🌟
💻 Reduce your agentic costs with the ELM-ITV Codebase
github.com/disler/elm-itv-benchmark
🔗 Links:
Bun https://bun.sh/
Ollama ollama.com
Promptfoo https://promptfoo.dev/
Apples OpenELM machinelearning.apple.com/research/openelm
📚 Chapters:
00:00 The cost of agentic tools is going to ZERO
00:48 Are ELMs ready for on device use?
02:28 Setting standards for ELMs
04:05 My (IndyDevDan) personal standards for ELMs
06:36 The ITV benchmark
07:05 ELM benchmark codebase
09:30 Bun, Ollama, Promptfoo, llama3, phi3, Gemma
12:10 Llama3, Phi3, Gemma, GPT3 TEST Results
16:10 New LLM class system
18:45 On Device PREDICTION
19:05 Make this prompt testing codebase your own
19:45 The cost of the prompt is going to ZERO
20:15 How do you know if ELMs are ready for your use case?
#promptengineering #aiagents #llama3
If you've found yourself stuck trying to control the flow of data between you and your AI Agents, this video is for you.
Two-way prompting is a powerful pattern that can help you build more useful and valuable agentic workflows. It involves a back-and-forth conversation between you and your AI agents to drive outcomes. Two-way prompts are essential multi-step human-in-the-loop interactions. They allow you to provide various types of feedback, such as selecting files, pasting URLs, and giving improvement suggestions to your personal AI assistant, AI Agents, and Agentic Workflows.
As the engineer, product builder, and user, you are the essential guiding force in your agentic workflows. Focus on driving results for yourself and your users, rather than just building agentic workflows for the sake of it.
Even if you're aiming to build 100% agentic workflows that work without your input, you can use two-way prompts (aka human in the loop) as stepping stones in your engineering while developing fully agentic workflows. Don't hesitate to create partial agentic workflows with human-in-the-loop interactions.
In this video you'll see firsthand how simple prompts can evolve into a dynamic conversation between you and your AI, turning complex tasks into a seamless dialogue.
Using v2 Ada, our proof of concept personal AI assistant written in python, we show you the incredible benefits of initiating a two-way, collaborative dialogue. We've added new agentic workflows to our LLM Router to utilize two-way prompting functionality. From scraping a URL to create usable code to refining a view component with real-time feedback, this video is packed with actionable insights. Using GPT4 Vision model, our AI Assistant is able to build and iterate on a Vue.js component. This idea can be expanded to work on any front-end like react, svelte, raw html, htmx, solid.js, and any front end framework. Using the two-way prompt, our personal ai assistant responds to our needs, making our engineering workflow more efficient and responsive.
Join us as we demonstrate two-way prompting concept with Ada, breaking down complex processes into simple, conversational steps.
Remember, these new AI patterns with AI Agents and agentic workflows are not about 'using AI' for the sake of it. They are about using AI to drive results for yourself and your users.
📖 Simon Willison’s files-to-prompt:
simonwillison.net/2024/Apr/8/files-to-prompt
🎥 Watch Ada v0 - Personal AI Assistant:
youtu.be/kLi4SKlc4HQ
🔬 Personal AI Assistant Gist (Draft/Incomplete v2):
gist.github.com/disler/1d926e312b2f46474b1773bace21f014#file-main9_ada_personal_ai_assistant_v02-py
📖 Chapters
00:00 A Pattern Hiding in Plain Sight
00:17 Two-Way Prompting
01:10 Personal AI Assistant Two-Way Prompt
05:00 Vue Component Agentic Workflow
08:20 Multi-Step Human In The Loop
09:04 Essential Insight - You are the asset
10:09 3 Tiers of Agentic Workflows
#aiagents #agentic #promptengineering
The answer is pretty clear, the best way to use your growing collection of AI Agents is in the form of a personal AI assistant.
Not just 'a' personal AI assistant, YOUR personal AI assistant.
Imagine a tool so powerful, it feels like an extension of your mind. In this video, we dive into the creation of the most important agentic application we can build and use: Your Personal AI assistant. This tool will be limited only by your imagination, and your ability to hop in your python or typescript code and COOK up great agentic workflows, AI Agents, prompt chains, and individual prompts. Your personal assistant can code for you, research for you, and organizing your digital life. BUT in order to get to that vision we have to take small, incremental steps. Here we look an EARLY prototype of what future personal AI assistants (the next level of VAs) will look like through ADA. ADA is the name of my, personal AI assistant. It's a prototype to show what this technology will be able to do for you.
In order to make use of your AI Agents and prompt chains in the form of your personal AI assistant, we need a framework for prompting your agents. In this video we introduce two critical frameworks for building your personal AI assistant: the PAR framework, and the simple keyword AI Agent Router (LLM Router). The PAR framework sets an a clean loop for you and your personal AI assistant. First you speak in natural language, we run text-to-speech (TTS) to capture your prompt and convert it into text which becomes your nlp/prompt (natural language prompt).
Next we use an LLM Router called the simple keyword AI Agent Router which takes your prompt and decides which AI Agents to run. Your agents run their individual, isolated workflows, and finally your personal ai assistant (ai va) responds to you using speech-to-text (STT) completed the PAR framework.
The beauty of this framework is that it doesn't make any assumptions about your prompts, prompt chains, or agents, all of that runs from the llm router based on your activation keywords based on your prompt. You can run langchain, crewai, autogen, or any other agent framework to build and run your agentic workflows. In future videos we'll be utilizing the AgentOS micro architecture to build reusable, composable AI Agents. Our LLM Router will then route to our individual agents to run dedicated functionality.
This is just the beginning of the most important agentic application we can build and use: Your Personal AI assistant.
Stay focused, and keep building.
📝 Personal AI Assistant Gist (Draft/Read Only)
gist.github.com/disler/2840f0404c44fc662f7673d783b89f81
🤖 AgentOS - Build reusable, composable agents
youtu.be/8wSH4XukcH8
🔍 7 Prompt Chains For Better AI Agents
youtu.be/QV6kaNFyoyQ
📖 Chapters
00:00 The Agentic Pieces are LINING up
00:33 Your Personal AI Assistant
00:55 ADA Demo, Proof of Concept
03:05 Big Ideas, PAR Framework, LLM Router, Flaws
03:50 Prompt, Agent, Response
06:25 AI Agent LLM Router
11:50 Future of Personal AI Assistants
13:15 Everything we do is STACKING up
14:07 Improvements, Flaws, vNext
17:05 More AI Agents, More Prompt Chains
#aiassistant #aiagents #virtualassistant
The LLM Ecosystem is ever evolving so in order to keep up, you'll need an architecture that has interchangeable parts that can be swapped in and out as needed. This is where the Agent OS comes in.
A great architecture, can future proof your AI agents and make them more adaptable.
The Agent OS is a micro architecture based off of Andrej Kaparthy's LLM OS. It's comprises three primary components: the Language Processing Unit (LPU), Input/Output (IO), and Random Access Memory (RAM). Each serves a unique purpose in the construction of AI agents, enabling you, the developer, to create systems that are not only efficient but also adaptable to the rapidly changing landscape of AI/LLM technology. The LPU, positioned at the core of the architecture, integrates model providers, individual models, prompts, and prompt chains into a cohesive unit. By storing all llm, and prompt related functionality into one component, the LPU, we can focus on prompt engineering and prompt testing around this unit of this Agent. This integration facilitates the creation of AI agents capable of solving specific problems with high precision. Thanks to the layered architecture, each piece can be swapped out. So when GPT-4.5 or GPT-5 rolls out, you can easily upgrade your AI agent without having to rebuild the entire system from scratch.
The RAM component enables your AI agent to operate on state, allowing it to adapt to changing inputs and produce novel results. The IO layer, on the other hand, provides the tools (function calling) necessary for your AI agent to interact with the real world. This includes making web requests, interacting with databases, and monitoring the agent's performance through spyware. By monitoring your AI agent's state, inputs, and outputs, you can identify issues and make improvements to the system.
In this video we dig into ideas of creating composable agents where the input of one agent can be the output of another agent. This is a powerful concept that can be used to create complex agents that can solve a wide range of problems. It's the evolution of the core idea agentic engineering is built on: The prompt is the new fundamental unit of programming and knowledge work. First you have llms, then prompts, then prompt chains, then AI Agents, and then Agentic Workflows. This is the future of programming and knowledge work.
🧠 Andrej Karpathy’s LLM OS
youtu.be/zjkBMFhNj_g?si=lY10VSHBUGDPA8Hs
🔗 7 Prompt Chains for Powerful AI Agents
youtu.be/QV6kaNFyoyQ
💻 Everything is a Function
youtu.be/q3Ld-MxlXmA
🔍 Multi Agent Spyware
youtu.be/UA6IVMDPuC8
📖 Chapters
00:00 Best way to build AI Agents?
00:39 Agent OS
01:58 Big Ideas (Summary)
02:48 Breakdown Agent OS: LPU, RAM, I/O
04:03 Language Processing Unit (LPU)
05:42 Is this over engineering?
07:30 Memory, Context, State (RAM)
08:20 Tools, Function Calling, Spyware (I/O)
10:22 How do you know your Architecture is good?
13:27 Agent Composability
16:40 What's missing from Agent OS?
18:53 The Prompt is the...
#aiagent #llm #architecture
Use these 7 Prompt Chains to build POWERFUL AI AGENTS with the help of Claude, Opus, Haiku or your favorite LLM.
The name of the game in software engineering is: How can I build agentic software where my AI Agents can do the heavy lifting for me? There are levels to this. You start with a single prompt, then you can chain prompts and code together to create powerful AI Agents that can do the heavy lifting for you. There are so many applications for this, from content creation to research to coding. Every single prompt chain is a potential 5,6,7 figure product. We're only scratching the surface with UIs like ChatGPT, Anthropic, Gemini and other Chat Based UIs. The future is bright for AI Agents and Agentic Applications.
Let's unlock the Prompt Chains that can enhance your prompt engineering abilities to elevate your software's capabilities. We're breaking down seven powerful prompt chains, complete with real-world examples, to show you exactly how to harness LLMs like Claude-3's Opus, Haiku, Sonnet, and whatever your favorite favorite LLM provider is. Discover how u create Agentic software that works tirelessly for you and your users, adding incredible value every step of the way.
The ideas we'll discuss are at the core of tools like Langchain, langgraph, Autogen, and CrewAI. While these tools are powerful, they're often overkill. Powerful AI Agents can be built simply by combining together several prompts in certain patterns and workflows. Call it prompt chaining, prompt orchestration, prompt graphs or whatever you like. From constructing compelling blog posts with the snowball prompt chain to building entire software modules via the worker pattern, this video is a goldmine for anyone looking to deploy AI in practical, impactful ways. Consider a free AI Prompt Engineering Course where we'll reveal several prompt orchestration patterns like the fallback prompt chain, a pattern than can save you time and money while ensuring your AI Agents are still reliable and effective.
It doesn't matter what you're building. AI Coding Assistants, Research Assistants, Personal AI Assistants, CLI Tools, all benefit from your ability to build prompts and your ability to compose prompts into useful patterns. Let's walk through seven distinct prompt chains, including the innovative snowball and worker patterns, showing you the path to automated content generation, sophisticated research tools, and even custom AI coding assistants. Discover how to make your software think, adapt, and solve problems with minimal input, unveiling a future where your software development process is as dynamic and intelligent as the market demands.
Composability, and Reusability are a big idea we focus on on the channel. AI Agents are no different. The more you can compose prompts together, the more powerful your AI Agents will be. The more you can reuse prompts, the more efficient your AI Agents will be. This is the future of software engineering. This is the future of AI Agents. This is the future of Agentic Applications.
✏️ Get These 7 Prompt Chains (Gist): gist.github.com/disler/409d9685c8b251ed723a7aca43cc4b9b
🗣️ When to use PROMPT CHAINS: youtu.be/UOcYsrnSNok
🐍 LLM Python Module: llm.datasette.io/en/stable/python-api.html#
🤖 LLM Claude Python Module: github.com/simonw/llm-claude-3
🛠️ How to Engineer Multi-Agent Tools: youtu.be/q3Ld-MxlXmA
🔮 2024 Predictions (AI, LLM, Coding, Agents): youtu.be/UES89QRc3Sk
📚 GPT Research (Worker Prompt Chain): github.com/assafelovic/gpt-researcher?tab=readme-ov-file
#aiagents #promptengineering #gpt
Ever imagined a world where your coding tasks could be completed with just a few keystrokes? Welcome to the future of programming with the best AI Coding Assistant. In this video, we dive deep into how AI-powered engineering is changing the game. We'll explore 3 One Shot Prompts For the best AI for Coding experience. We'll utilize Cursor AI, a popular AI coding assistant, to demonstrate how you can code faster and more efficiently. Keep in mind though, these one shot prompt engineering techniques can be used with any AI coding assistant. I often times go back and fourth between Cursor AI and Aider, all the ideas translate well between the two. Because it's really all about the prompt. Is Cursor the best AI for coding? We'll investigate that in another video. This is all about improving our prompting ability with 3 incredible one shot prompts.
The prompt is the new fundamental unit of programming.
By utilizing concise prompts with your AI Coding Assistant you can generate your python, javascript, typescript, and whatever language you're using, faster than ever. You will unlock lightning-fast coding speeds without sacrificing quality. Experience firsthand the power of prompt-based coding as we navigate through several coding challenges using AI in a production code base. Learn how to effortlessly add functionality, refactor code, and more with just a handful of well-chosen words. Coding has never been this efficient, or this accessible.
We'll start off with basic GPT-4 prompts but then we'll use the new claude-3-opus model to the test with a more complex prompt that will generate multiple functions for us in a SINGLE ONE SHOT PROMPT. We'll showcase why Claude 3 Opus is the best model.
This isn't just about coding faster—it's about thinking differently. Learn how the new wave of 'agentic engineers' leverage AI tools to not only code but also to design and ideate more effectively. We'll take you through real-life examples of how prompt keywords can drastically reduce coding time and amplify your creative potential.
These are just a few of the core ideas we'll have in our prompt engineering course. Coming soon.
We're about to blast through 10k subscribers, so if you're new here, hit that subscribe button and join the community. We're all about helping you become a more effective developer.
Stay focused, keep building.
🚀 Learn More AI Coding Assistant Techniques
youtu.be/Smklr44N8QU
🧠 What Is a True AI Coding Assistant
gist.github.com/disler/20ae1bf472dbe5b743a0161a9da42a42
👨💻 Cursor - AI Coding Assistant
https://cursor.sh/
📖 Chapters
00:00 AI Coding Assistant One Shot Prompts
01:00 Only Coding Assistants gives you this SPEED
01:50 Coding Prompt With File Context
02:35 Use this technique to generate stateful code
03:50 AI Coding helps you write readable code
05:00 THANK YOU - ALMOST AT 10k SUBS
06:30 The most important one shot prompt keyword
09:50 Claude 3 Opus The Best Model For AI Coding
12:30 There's a ton of value in concise prompting
13:40 Your coding assistant is much more powerful than you think
#aiassistant #aicoding #promptengineering
With technology's INSANE pace, the threat of AI software engineers like Devin looms large. But what if this isn't the END GAME, what if this is a critical pivot point for your career?
Let's peels back the layers of fear surrounding AI and LLMs to reveal the core strategies for not only coexisting with but commanding these new AI tools. Unearth the untapped potential of AI assistants like Devin, Cursor, Aider, and whatever comes next to transform them from adversaries to powerful assets that lift your career to new heights. It's clear that GPT powered tools are not going away and the model provider like OpenAI, Mistral, Anthropic, Google and others are only going to get better. That means the future of software engineering will require us to master Prompt engineering, LLMs, and the tools they enable.
In the fast-evolving landscape of technology, being merely good isn't enough; being indispensable is what defines success. This video takes you beyond the basics of surviving the AI revolution in software engineering, presenting a compelling case for becoming the master of new AI tooling, LLMs, and AI coding assistants Cursor and Aider and now, AI Software Engineers like Devin. Learn the secrets to not just staying relevant, but to completely winning in the age of AI.
There are three pivotal areas that define your foothold in the tech industry: your A to L (asset to liability) ratio, the relentless evolution of your engineering toolkit, and the mastery of AI tools to ensure you're not just surviving, but thriving. Learn how to position yourself as an indispensable asset, with insights into leveraging AI as a powerful ally in your career journey.
Shift your mindset and skillset to what's required for the age of AI. From understanding your asset to liability ratio to embracing the continual evolution of your engineering toolbox—this video lights the path. With practical examples and step-by-step guidance, learn how to leverage AI tools like Cursor and Aider to safeguard and propel your career forward.
🐦 Andrej Karpathy Tweets On Automating Software Engineering (Calls out our Copilot++ Video - so cool)
twitter.com/karpathy/status/1767598414945292695
▶️ Cursors Copilot++ vs GitHub Copilot Video
youtube.com/watch?v=Smklr44N8QU
🤖 Introducing Devin
cognition-labs.com/introducing-devin
📄 Microsoft's AutoDev Paper
arxiv.org/abs/2403.08299
📖 Chapters
00:00 Fear Power Control
00:30 Big Ideas
01:30 Do Companies Need Software Engineers?
03:00 The Kings Of Knowledge Work
04:28 Entrepreneur, VC, Investor's Mindset
04:53 "Do we still need software engineers?"
06:13 The New Kings
08:20 Evolve, Thrive, Die
09:56 The New Interview Question
11:34 Use your fear of Devin to learn Devin
12:00 Recap for Job Security and Career Growth
13:25 OpenAI Engineer Tweets Our Video
14:26 It's all about evolving your AI toolset
14:56 Take Action - Learn AI Coding Assistants Now
#layoffs #aiagents #copilot
Ever wondered how leading language models stack up in a head-to-head challenge? In our latest video, we simplify the complex world of LLM benchmarks with a BATTLE ROYALE that pits Grok, Vertex Pro, GPT-3.5 Turbo, and Claude Sonnet against each other. Get ready for an interesting showdown that will change the way you look at language models.
The rules are simple...
LAST LLM STANDING WINS
Join us as we show off a unique benchmarking tool built in a couple hours thanks to AI Coding Assistants like Aider and Cursor. Last LLM Standing Wins is a simple LLM benchmarking tool where you utilize prompt testing framework promptfoo to generate your tests and then visualize how they perform in a deterministic visual way. We setup several battles where Grok, Vertex Pro, GPT-3.5 Turbo, and Claude Sonnet battle it out over a series of prompt testing challenges surrounding NLQ to SQL prompt tests. Experience the thrill of live competition and gain insights like never before.
Step into the arena with us as we handpick top language models to face off for speed, accuracy and cost. 'Last LLM Standing Wins' benchmark makes it easy to compare the performance of different models, revealing their strengths and weaknesses. The best part is that it's built ontop of the prompt testing framework promptfoo, which is a testing driven prompt engineering framework that allows you to test your prompts and truly KNOW if your LLM is performing well or not for your specific use case.
The battlefield is set, and the contenders are ready. With each model running the same tests, we meticulously evaluate their performance, uncovering strengths and weaknesses. Grok's LPU mixtral model FLASHES through the challenges, bagging the frugal award for least cost, while others vie for the bullseye award for error-free executions. We witness Anthropics Claude models struggling with rate limits and GPT-3.5 Turbo's impressive balance between speed and accuracy.
The suspense is real as we tally the scores, revealing which LLMs stand tall and which crumble under the pressure. By applying a transparent, simple and easy-to-understand benchmarking tool, we shine a light on the performance of each model, bestowing awards for efficiency and accuracy. Every moment is packed with action and enlightening insights, making it impossible to look away.
The prompt is the new fundamental unit of knowledge work and programming.
Master the prompt, and you master the LLM.
🛠️ Prompt Testing
https://promptfoo.dev/
🧪 Testing Driven Prompt Engineering
youtu.be/KhINc5XwhKs
🆚 Gemini Pro vs GPT 3.5 turbo LLM Benchmarks
youtu.be/KhINc5XwhKs
💬 Text to SQL to Results
talktoyourdatabase.com
🏁 Claude Benchmarks
anthropic.com/news/claude-3-family
📖 Chapters
00:00 LLM Benchmarks are biased and complex
00:36 Last LLM Standing Wins
02:20 GPT-4 vs Claude Opus
03:44 Promptfoo Testing Framework
05:44 Speedster Test - GPT-3.5 Turbo vs GROQ
09:00 Final Mega LLM Battle
10:50 Groq LPU Mixtral is insanely fast
12:20 Benchmark your personal prompts
13:22 A real production use case
15:10 KNOW that your prompts are working
#llmbenchmark #bestllm #openai
The days of manually writing out every line of code is far over. If you're still typing every line - this is a wake up call for you. I don't want you to get left behind in the wave of LLM powered AI coding tools.
Keeping your eye on AI Coding Assistants is critical to staying ahead of the curve in the engineering world where layoff rates are high and the competition is fierce.
Let's talk Cursor Copilot++ vs Github Copilot...
In this video, we will compare Cursor Copilot++ and Github Copilot, two of the most popular AI coding assistants on the market.
Right away though we have to make a differentiation between the two. Cursor, just like Aider, is a TRUE AI Coding assistant. It is a tool that is designed to help you code better, faster, and more efficiently. While the base version of Github Copilot is a code completion tool.
Cursor's new Copilot++ aims to redefine coding efficiency, allowing for simultaneous auto-completions and context-aware suggestions that will leave you in awe. This video is a testament to how AI can not only enhance but revolutionize the way we code. By enabling features like multi-line auto-completion and context-aware suggestions, Copilot++ is setting new benchmarks in coding efficiency.
AI Coding Assistants like Aider are two of the best AI for coding, enabling you to think LESS about individual lines of code. They enable you to think about the bigger picture, and give you the ability to up level your perspective. Think more like a product manager, or UX designer, and use prompts to handle the more mundane lines of code.
Let's walk through a solid front-end example using Vue.js, demonstrating live how Copilot++ seamlessly integrates into your coding projects, and drastically reduces your coding time while increasing accuracy. See it in action as we tackle real coding challenges, showcasing how every command leads to precision-driven results.
Witness firsthand the magic of Copilot++ as we showcase its ability to handle multiple coding prompts simultaneously, from adding and removing divs and handling logic with high accuracy to crafting complex coding functions effortlessly. This video is your ticket to mastering Copilot++ and supercharge your AI coding and AI engineering abilities with LLM powered AI coding assistance aka AI Coworkers.
💻 Cursor Copilot++
https://cursor.sh/cpp
🔗 Vue Flow
https://vueflow.dev/
🤖 What makes a TRUE AI Coding Assistant (TACA)?
gist.github.com/disler/20ae1bf472dbe5b743a0161a9da42a42
📖 Chapters
0:00 One of the BEST coding assistants
0:25 Multiple Prompts - In Parallel
1:00 Copilot++
4:00 VueFlow
5:00 Iterative Prompting
6:25 Copilot++ Multiline Completion
7:45 Adding Web Documents To Rag Context
10:35 Copilot++ Function Completion
11:30 Copilot++ Intelligent Auto Rename
12:20 Copilot++ Helping us again
13:00 AI Coding Takeaways
#githubcopilot #copilot #aicodingassistant
Discover how to leverage AI coding assistants to turbocharge your productivity in one critical aspect of software engineering: WRITING DOCUMENTATION.
A hidden use-case of AI coding assistants like Aider and Cursor is their ability to write documentation for you based on your code, and in this video, we'll show you how to do it. We'll also demonstrate this by looking at a real application we're building on the channel: the IndyDevTools, a toolkit built for developers to solve problems in a principle driven, reusable way using LLMs. We'll show you how to use AI to write documentation for your code, and how to keep track of your prompts so you can reuse them in the future.
AI Coding is taking off and we're inside the rocket ship.
By using AI Coding Assistants to generating documentation you can ship code faster, you can communicate the value and use of your code, feature and product. We also show off the a new tool in the IndyDevTools called the Simple Prompt System, which is a tool that helps you keep track of your prompts and reuse them in the future. This two for one video is a must watch for any developer looking to level up their productivity and understanding of AI coding assistants.
This is all about your building your ability to use AI to write documentation for your engineering, and this is all about templating your prompts so you can reuse them in the future.
Focus on the signal, not the noise.
See you in the next one.
🤖 AI Coding Assistant
https://aider.chat/
🛠️ Build your REUSABLE PROMPT SYSTEM
github.com/disler/indydevtools
#aicoding #aiengineering #promptengineering
The AI landscape is evolving at an unimaginable pace, and staying ahead requires not just insight but a principled, forward-thinking approach. In our latest exploration, we dive into the extraordinary feats accomplished by Gemini Pro 1.5 and Sora, deciphering how these breakthroughs signal a shift in the way we develop, interact, and use LLM technology.
The announcement of Google's Gemini Pro 1.5 and OpenAI's Sora, the ground breaking text to video model, represents more than just technological advancements; they symbolize a tidal wave of changes set to redefine the landscape of AI Agents. With unique insights and a focus on what this means on both a macro and micro level, in this video aims to prepare you for the impact of these innovations.
As engineers and product builders, the rapid acceleration of these technologies presents both exhilarating opportunities and formidable challenges. Our LLM Agents, AI Copilots and AI Powered Tools can do more than ever before, and the pace is NOT slowing down. So how do we harness this incredible power? Which tools should we adopt to stay ahead in the game? This video dives deep into what Gemini Pro 1.5 and Sora mean for us, offering actionable insights and a toolkit designed to keep you at the cutting edge.
Facing the wave of AI advancements head-on, we delve into the critical decisions facing today's engineers and developers: choosing the right technologies, building adaptive tools, and focusing our attention where it truly matters. Join us as we outline a agentic toolkit, indydevtools, built from our comprehensive three-part video series, designed to reveal patterns and principles for building multi-agent apps to ensure you remain at the forefront of this technological tsunami. We'll discuss the implications of these developments, from the way we prompt engineer to how we approach the very act of innovation, ensuring you have the tools and knowledge to build incredible AI Agents to help us thrive in this new ever expanding era of AI.
🌌 OpenAI Sora, Infinite Content Generator
openai.com/sora
🔮 Gemini 1.5 Pro: 1 Million Token Context Window
https://blog.google/technology/ai/google-gemini-next-generation-model-february-2024/#performance
🛠️ IndyDevTools
github.com/disler/indydevtools
📄 Attention Is All You Need
arxiv.org/abs/1706.03762
📚 Chapters
00:00 Keeping pace with Gemini, Sora, and what's next
02:00 Gemini 1.5 Pro - INFINITE context
09:46 Sora - INFINITE content
15:00 Micro - Principles to keep up - IndyDevTools
#LLMAgents #YouTubeAutomation #GeminiPro
In this video we use the Apple Vision Pro to DISCOVER how AI agents and LLM technologies are POWERING the next wave of YOUTUBE content creation.
Imagine enhancing your content creation workflow with the cutting-edge capabilities of AI coding and multi-agent systems. Our latest video unveils how you can use multi-agent principles with a practical example: YouTube metadata automation, deeply integrated with LLM agents and AI technologies. We utilize python and poetry as our language and package manager of choice. This tool is not just about streamlining your upload process; it's about leveraging the advanced features of AI agents to transform the cumbersome task of generating metadata for any type of digital assets. In this video we create the Proof Of Concept of your LLM Powered YouTube metadata generation tool in a modular, reusable fashion.
Dive into an in-depth exploration of this tool and the application of principles for building agentic applications. We utilize the Apple Vision Pro to create a new generation of youtube developer content. By combining the concepts of agentic engineering and AI coding, we demonstrate how YouTube automation becomes not just a possibility but a practical tool in your creative arsenal fueled by your LLM agents. This video goes beyond a simple Apple Vision Pro review; it's a comprehensive guide that sits in the intersect of AR/VR with the Apple Vision Pro, LLM Agents, and applying engineering principles to automate and enhance your YouTube content strategy, offering a glimpse into a future where creators can focus more on creativity and less on manual tasks.
🔗 Part 1: Principles for building Multi-Agent Applications
youtu.be/q3Ld-MxlXmA
⏰ Stop wasting time writing SQL
talktoyourdatabase.com
⚙️ Faster-Whisper - Transcribe Audio/Video FAST
github.com/SYSTRAN/faster-whisper
In this video we discuss core engineering principles and techniques essential for designing and coding sophisticated multi-agent systems. Discover the key to rapid engineering and productivity gain with a principle so potent that experienced senior engineers naturally learn it instinctively. "Everything is a function" isn't just a phrase; it's a transformative approach that will revolutionize how you architect, design, code, and deploy software, making the creation of AI agents and high-quality applications more efficient than ever.
Embark on a journey to automate the exhaustive process of YouTube content creation, from SEO research to titles, to thumbnails to descriptions to chapters. In just 3 videos, we'll automate the entire process. All you'll need is your rendered video. This series will guide you through building a YouTube automation metadata generator, showcasing how to leverage agentic engineering and LLM technology to streamline content creation. The how is more important than the what. Focus on the principles and the underlying strategies to replicate this solution to anyone of your applications. By breaking down complex problems into manageable tasks and employing AI agents judiciously, you'll learn to maximize efficiency and minimize manual effort. Whether you're a seasoned software engineer or an aspiring developer, this video offers practical insights and a solid foundation in leveraging AI technology for software development. Let's push our engineering into the future by practicing employing AI Agents to automate our YouTube Metadata content creation.
I forget to mention it in the video but AI Coding Assistant tutorials are IN DEVELOPMENT! Super excited to share those with you and eventually lead the tutorials into full blown courses to push your engineering into the future of programming. Stay tuned for that. The initial set of tutorials will be released right here on the channel.
🔗 2024 Predictions for AI & LLMs
youtu.be/UES89QRc3Sk
⚙️ Text to SQL
talktoyourdatabase.com
#youtubeautomation #aiagents #gpt
Our video takes a practical approach to understanding the viability of on-device local LLMs. We not only discuss their current limitations in terms of speed, RAM, and GPU requirements but also demonstrate how to effectively test these models using the incredible prompt testing tool, Promptfoo. By comparing local models like Mistral, PHI-2, and Rocket 3B against cloud counterparts in various scenarios, including a natural language query to SQL, we provide a clear picture of where local LLMs stand today. Our hands-on tests reveal the real-world performance of these models, underscoring their potential and pitfalls. We also explore innovative solutions like LlamaFile for running LLMs locally, offering viewers a glimpse into the future of AI technology. This video is an essential guide for developers and tech enthusiasts looking to stay ahead in the rapidly evolving world of local language models. Join us as we unravel the complexities of on-device LLMs and their journey towards becoming a practical, everyday reality.
💻 Prompt Testing Codebase
github.com/disler/llm-prompt-testing-quick-start
🔗 Links
Promptfoo: https://promptfoo.dev/docs/intro
LLamafile: github.com/Mozilla-Ocho/llamafile?tab=readme-ov-file
Llamafile Quick Start: github.com/disler/lllm
Mistral: mistral.ai
💻 Learn Prompt Testing
youtu.be/KhINc5XwhKs
🤖 Try Llamafile for local LLMs
youtu.be/XnoKvdeZAN8
⭐️ Text to SQL to RESULTS
talktoyourdatabase.com
#llama #llm #googlegemini
Ever wondered why AI coding assistants, despite their groundbreaking potential, haven't taken the engineering world by storm? This video delves into the heart of this enigma. While tools like Aider, Cursor, AutoGen, and CrewAI aim to revolutionize coding by potentially 5x'ing productivity, their journey to mainstream success is hindered by significant, yet solvable, challenges that they all aim to overcome.
We explore the critical barriers blocking the widespread adoption of AI pair programming - from file management and accuracy issues to speed and security concerns. Moreover, we confront an overlooked yet significant issue: Skill Atrophy due to over-reliance on AI assistance. By dissecting these problems, we not only understand why rapid prototyping coding LLMs tools like GPT Engineer and ChatDev can't be considered true AI Coding Assistants but also offer a standard for what makes up TRUE AI CODING ASSISTANTS. We'll discuss actionable solutions and strategies to overcome existing problems with Ai Coding Assistants.
This video is a treasure trove for Jr.+ engineers and product enthusiasts eager to harness the full power of AI Copilots in programming. It's not just about identifying the problems - it's about unlocking the vast potential of AI coding assistants to transform your productivity and push the boundaries of what's possible in coding. If you're ready to be at the forefront of the AI revolution in coding, this is the video for you.
🔗 AI Coding Resources
- Aider: https://aider.chat/
- Cursor: https://cursor.sh/
- Standards for AI Coding Assistants: gist.github.com/disler/20ae1bf472dbe5b743a0161a9da42a42
🤖 2024 Predictions for LLMs and AI Coding Assistants
youtu.be/UES89QRc3Sk
💻 Stop wasting time writing SQL
talktoyourdatabase.com
We navigate through the macro and micro strategies of the GPT Store. From a macro perspective, OpenAI is clear in its direction - leading the AI race, building an ecosystem accessible to ALL, and taking smart, incremental steps in the Generative AI space. The GPT Store differentiates itself significantly the Apple App stores by making AI creation accessible to non-engineers, expanding the total addressable market. This move, however, brings both opportunities and challenges, including the need for more refined filtering mechanisms due to the potential influx of low-quality GPTs.
On the micro side, we focus on actionable steps for engineers. How can you, as an engineer, leverage this platform? Should you create your own GPTs or wait for the dust to settle? The key lies in understanding the true value proposition of GPTs: domain-specific knowledge bases, message threads with contextual understanding, and the untapped power of API actions. We also explore the potential of applications like a text-to-SQL converter (Talk To Your Database) and how such innovations can integrate with GPTs to revolutionize workflows.
Explore these exciting developments in AI and engineering. We're here to guide you through the macro and micro of the GPT Store and how it can elevate your engineering skills and products. Don't forget to share your thoughts in the comments, and if this content resonates with you, like, subscribe, and stay tuned for more insights on AI-driven engineering.
🔗 OpenAI GPTs Store Announcement:
openai.com/blog/introducing-the-gpt-store
🚀 Text To SQL To Results:
talktoyourdatabase.com
📖 Chapters:
00:00 Is the GPT Store worth your time?
00:35 GPTs are a rough first draft
01:23 Macro strategy of the GPT Store
01:55 OpenAI App Store vs. Apple App Store
03:45 True engineers care about their craft
05:13 UGC on AGC
07:40 Micro strategy for engineers
09:15 Value proposition of GPTs
10:15 Custom Actions Real Example - Text to SQL
12:30 What do you think about the GPT Store?
13:30 Engineers have a unique advantage
#gpts #openai #sql
I figured this would be a great opportunity to show off the future of AI-powered coding demonstrating how you can supercharge your software engineering skills using Copilots like Aider, and Cursor and tech like Electron and DuckDB. In our AI Devlog walkthrough, we break down how to tackle the min, mean, and max calculation challenge on a massive dataset using DuckDB. We'll use the latest version of turbo4, an OpenAI assistants API wrapper that enables you to build knowledge bases directly from URLs. Your custom turbo4 assistant can then consume the knowledge base and generate code for you.
This channel is dedicated to helping you transform from a traditional coder to an Agentic engineer, adept in leveraging the potency of AI-enhanced coding techniques and next-gen tools like the Assistants API from OpenAI to elevate your productivity and capability in the modern coding landscape.
We start by setting up an Electron app and guide you through using Typescript, the Vuetify framework for Vue UI components, and DuckDB's in-memory database for handling ONE BILLION rows of data efficiently. Our focus is on clear, applied knowledge, we don't care about typescript types we care about learning how to build like we're an engineer of the future. We'll utilize Two AI Coding assistants: Aider and Cursor. We'll talk and show what they're both good at and where they're weak. By following our methodical approach, you'll learn valuable strategies for improving prompt engineering abilities and get ahead of the AI wave that's redefining software engineering. Subscribe to our channel for more professional insights into AI coding, and check out our description for links to Aider, Cursor, and the Assistants API documentation.
Massive shout out to all the engineers who have participated in the One BILLION Row Challenge and all the engineers that's code has been used as a knowledge base to build large language models.
💻 One Billion Row Challenge - Electron Edition
github.com/disler/1brc-electron
🌎 2024 Predictions For AI & LLM Engineers
youtu.be/UES89QRc3Sk
🔗 Resources
1 BRC Original - https://www.morling.dev/blog/one-billion-row-challenge/
1 BRC DuckDB Post - rmoff.net/2024/01/03/1%EF%B8%8F%E2%83%A3%EF%B8%8F-1brc-in-sql-with-duckdb
DuckDB - duckdb.org
Aider - https://aider.chat/
Cursor - https://cursor.sh/
Electron Vite Vue Typescript Starter - github.com/Deluze/electron-vue-template
Vuetify Server Table - vuetifyjs.com/en/components/data-tables/server-side-tables/#examples
Vuetify Pagination - vuetifyjs.com/en/components/paginations/#disabled
Electron - electronjs.org
LLM In CLI - github.com/simonw/llm
📖 Chapters
00:00 - One Billion Row Challenge
01:28 - Let Cursor Code For You
04:25 - Let Aider Code For You
10:44 - Electron IPC
14:30 - Generating (not quite) 1 Billion Rows
15:30 - OpenAI Assistants API Via Turbo4
19:00 - Build Micro Knowledge Bases
24:00 - Agent DuckDB SQL & Typescript Generation
27:20 - SWEET DuckDB Commands
29:26 - Cleaning up pageTable.ts
31:05 - End to End Electron App
33:40 - AI Coding the Frontend
44:47 - One Billion Rows In Electron
49:00 - Recap & Big Picture Agentic Engineering
53:55 - Talk To Your Database (text to sql to results)
#promptengineering #aider #copilot
Have you ever wondered why we all decide to make predictions?
Great Predictions yields
Greater Accountability yields
More Skin in the game yields
GREAT OUTCOMES!
Engineers, it's time to gear up for a gnarly 2024. This video zeroes in on Advanced AI and LLM trends. brimming with insights for both novice and veteran engineers. These 2024 predictions are hyper focused on prompts and prompt engineering as THE new fundamental unit of technology and programming. In this video we'll make predictions on AI & LLMs, Programming, Big Tech, Stocks & Crypto, Job Market, and Pareto Actions. The point isn't for my to flex my opinion, it's about sharing my perspective and getting you to think about your own predictions for 2024. I'm not a financial advisor, but I do have a track record of making accurate predictions.
With a scoreboard of 66% on 2023's predictions, we're primed to dissect the driving forces behind AI, programming paradigm shifts, and the tech industry's tides. Advanced AI and LLMs are completely altering how we think about, plan and build software. The industry as a whole is changing, from beginner to the senior+ level. In this video we'll predict OpenAI's dominance, open-source LLM breakthroughs, and touch on AI safety as an ongoing concern. It's become clear that prompt engineering will emerge as a critical skill, transforming the essence of knowledge work and how we interact with technology. Shift your focus: Big tech predictions, stock market insights, and crucial job market shifts are dissected with a straightforward, no-nonsense perspective aimed at providing the best experiences for you to make better decisions throughout 2024.
🔗 Resources
2024 Predictions Blog: indydevdan.com/blogs/2024-predictions
2023 Prediction Results: indydevdan.com/blogs/2023-prediction-results
📖 Chapters
00:00 Hey Engineers, Happy New Years!
00:45 Why Make Predictions?
02:13 2023 Predictions Track Record
03:29 AI & ML Predictions
06:30 Programming Predictions
11:28 Big Tech Predictions
15:00 Stock & Crypto Predictions
17:02 Job Market Predictions
19:27 Pareto Actions - 80% of the results with 20% of the effort
23:54 What are your predictions for 2024?
#2024predictions #promptengineering #programming
Financial Disclaimer: This video is for educational and informational purposes only and should not be considered as professional financial advice. The content represents personal opinions and may not reflect the most current or accurate financial information. Investing involves risks, including potential loss of principal. Viewers are advised to consult with qualified financial professionals before making any investment decisions.
I've been blowing off local llms since the beginning.
"It's too slow"
"They're to hard to run locally"
"Accuracy is too low"
There WERE many reasons to avoid local LLMs but things are changing.
I'm really excited to say llamafile and advancements in local LLM development is rapidly changing my perspective on local LLMs.
With just ONE line of code we can now run local llms. Thanks to Llamafile, we can now run local large language models (LLMs) with unprecedented simplicity. In this new devlog, we spotlight Llamafile's revolutionary single-command execution for local LLMs, transforming open-source AI accessibility for developers and engineers alike. Discover how you can set up and run local models like Mistral 7b Instruct and Facebook’s Wizard Coder effortlessly, while also learning to establish a reusable bash function for on-the-fly execution of any local Llamafile within your terminal.
Don't get me wrong, local LLMs are still not perfect. They are still lacking hard on key LLM benchmarks and the accuracy hangs low but it's not about where they are it's about where they will be. They are rapidly improving and soon, with proper prompt testing, they'll be viable to solve problems. Thanks to llamafile they are also getting easier to run locally.
Stay ahead in the fast-evolving world of AI with local models that are fast and open-source, made possible by Llamafile. This devlog not only showcases the astonishing ease of initiating local LLMs but also pays credit where it's due to appreciate to Justine's insane coding abilities (she wrote llamafile and cosmopolitan 🤯). We're diving deep into the synergy between stellar engineering and the democratization of AI technology. By the end of this video, you'll be well-equipped to integrate Llamafile into your workflow, enhancing your AI coding projects with the robust capabilities of local models and preparing you for whatever is next for local open source models. Subscribe to stay updated on the latest in AI devlogs, and make sure to like and share for more content on AiDER, local LLMs, and leveraging Llamafile for your development needs.
🚀 local llms - llamafile quick start
github.com/disler/lllm
💻 Incredible Resources
LLAMAFILE codebase --- github.com/Mozilla-Ocho/llamafile/tree/0.3
Core author --- creator of llamafile & cosmopolitan libc: https://justine.lol/
Original Blog Post --- https://justine.lol/oneliners/
Original llamafile introduction --- hacks.mozilla.org/2023/11/introducing-llamafile
How llamafile works --- github.com/Mozilla-Ocho/llamafile/tree/0.3?tab=readme-ov-file#how-llamafile-works
📖 Chapters
00:00 Llamafile
01:24 Local llm in 1 minute
02:24 Done - this is incredible
03:55 Run Local LLM Web Server UI
06:50 lllm - Prompt Engineering Aider
07:36 Aider
09:00 lllm - local large language models
12:11 Add Wizard Coder With AIDER
12:53 Wizard Coder via llama file
16:12 lllm - reusable local model bash function
16:47 Prompt - Why use local open source models?
#llm #llama #promptengineering
In this in-depth evaluation, we unpack the subtleties between the two second tier, SPEED dominating AI models: Gemini Pro and GPT-3.5 Turbo. Unlike most comparisons, this video draws on real-world testing and the latest industry insights and real benchmarking that speaks directly to senior engineers, and product builders. We reveal the nuances in LLM performance, uncovering slight edges and significant pitfalls across various metrics including speed, language understanding, instructional adherence, and developer experience.
Let's dive deep into a legit analysis based using the incredible testing frameworks Promptfoo. Price, multimodal capabilities, API complexity, AI alignment, and biases - every aspect of Gemini Pro and GPT-3.5 Turbo is revealed to give you insight in to rather or not you should be paying attention to gemini LLM tech.
🔗 Developer Resources
LLM Prompt Testing Quick Start - github.com/disler/llm-prompt-testing-quick-start/tree/v2
Promptfoo - https://promptfoo.dev/
Promptfoo Vertex AI - https://promptfoo.dev/docs/providers/vertex
Comparison Blog - klu.ai/blog/gemini-pro-vs-gpt-3-5-turbo
Gemini - https://deepmind.google/technologies/gemini/#introduction
Talk To Your Database - talktoyourdatabase.com (7777)
📘 Chapters
00:03 No Time Wasted, Should You Use Gemini Pro?
01:00 Gemini Pro vs GPT-3.5 Highlights
05:43 My Top Priority LLM Benchmarks
07:23 Gemini Pro Prompt Testing with promptfoo
11:30 Second Test: Gemini Pro vs GPT-3.5 Turbo
13:30 Prompt Testing is REALLY important
14:35 35 Success 4 Failed Test Cases
14:55 In this test, Gemini Pro's speed blows GPT-3.5 out of the water
16:30 Prompt Testing Helps You Build Better Products
17:07 Let's let the hype cycle die
18:20 Tweak all these tests to your liking
18:40 Regex prompt assertion match
19:20 Promptfoo is great for prompt evaluation and testing
20:24 Why test your prompts?
22:14 Competition means price goes down benchmarks go up
#promptengineering #googlegemini #gpt
What might surprise you is how simple prompt testing can be (shout out to the promptfoo developers). Promptfoo will enhance your prompt engineering skills and AI Agents with simple yet customizable LLM testing and evaluation. Promptfoo even has support for testing the new OpenAI Assistants API! It doesn't matter if you're using AutoGen, Assistants API, ollama, ChatDev, Aider, custom agents, multi agent systems or really any other prompt engineering tool. At the end of the day every tool is driven by prompts and that means llm testing and evaluations will help you gain confidence, cut costs, and optimize the results from your prompts.
This tutorial provides a hands-on approach to understanding the intricacies of prompt comparison and optimization. You'll learn token usage, time to completion and how to compare different prompts to choose THE WINNER. Promptfoo enables you to effectively compare and select LLMs, with a focus on achieving the best balance between speed, accuracy, and cost. We discuss key strategies for testing prompts in various scenarios, highlighting the importance of prompt evaluation and testing by looking at real llm test cases using GPT-4 Turbo and GPT-3 Turbo.
Let me know if you're interested in more prompt testing tutorials, frameworks, and methodologies.
📺 Quick Start LLM Testing
github.com/disler/llm-prompt-testing-quick-start
🔗 LINKS:
Promptfoo: https://promptfoo.dev/
Talk To Your Database: talktoyourdatabase.com (Beta Passcode 7777)
📖 CHAPTERS:
00:00 Are your prompts even good?
00:45 For real apps, prompt FEEL is not enough.
01:14 Cheaper, Faster, Accurate Prompts with PROMPTFOO
01:35 Quick Start LLM Testing
03:20 Immediate Results with GPT-4-Turbo vs GPT-3-Turbo
04:00 Clean and Reusable testing structures
06:00 Asserts and Test Cases
07:45 Learn these 3 components and you're good to go
09:35 LLM Evaluation and Testing 2nd Run
10:00 GPT-4 is breaking the bank and our timeline
12:33 Promptfoo has a lot more to offer for llm testing
13:12 Promptfoo has a OpenAI, Anthropic, Ollama, and soon Gemini Providers
13:30 Three reasons you should test your prompts
14:42 Test Driven Prompts
💬 Hashtags
#gpt #promptengineering #aiagents
This video marks the END of a series where we explored the comprehensive process of building an AI agent from the ground up to analyze our data. Understanding the "how" is just as vital as knowing the "what". We built a Multi-Agent Postgres Data analytics tool with new LLM based technologies like GPT-3.5, GPT-4, GPT-4 Turbo, AIDER, AutoGen, Guidance, Assistants API and more. We utilized existing and new prompt engineering to craft our more than 5 versions of our multi-agents data analytics tool. We dove deep in to multi-agent conversation flows, orchestration, memory management, chat monitoring and more. We explored the future of programming and how we can leverage AI to accelerate our ability to generate valuable products for both ourselves and the organizations we collaborate with.
With every video, every commit and every line of code it's become clear that the landscape of software engineering is changing dramatically. Those that embrace the change will be the ones who thrive in the future. Those that don't will be left behind.
I want to thank everyone at OpenAI for development such incredible technology.
I also want to thank everyone who built every tool we used throughout this series.
You are pushing the world into the future and I'm grateful to be a part of it.
On this channel, we don't just talk - we think, plan, and build. With the ultimate goal of PULLING the future of programming into the present to accelerate our ability to generate valuable products for both ourselves and the organizations we collaborate with.
Thank you for watching and I hope you enjoyed and learned from this series as much as I did.
🚀 TTYDB - Talk To Your Database
talktoyourdatabase.com
or
talktoyourdb.com
🤖 The Postgres AI Data Analytics Series
youtube.com/playlist?list=PLS_o2ayVCKvDzj2YxeFqMq9UbR1PkPEh0
💻 The Codebase
github.com/disler/multi-agent-postgres-data-analytics/tree/v10-talk-to-your-database-beta-launch
Youtube Exclusive Beta Launch Code: 9999
📘 Chapters
00:00 The Way We Interact With Data Is Changing
00:30 Great Questions Yield Great Outcomes
01:15 Talk To Your Database
02:15 Hyper Minimal UI
03:10 Our first AI Query
04:05 Insanely concise NLQ - users join job limit 10
05:40 Let's do some nlq data analytics
07:12 Copy SQL and Download at anytime
07:30 Let's be real - ttydb is not perfect
08:45 Great way to quickly explore data
09:30 All the value of SQL without the SQL
10:16 Four words - 10x your database interactions
12:30 Let's talk about SECURITY
14:50 Customize your TTYDB Experience
15:35 Automatic data expirations
16:13 We keep your data safe but not having it
16:28 Integrated Community Development
17:23 Don't be spammy, annoying or angry - TTYDB AI agents will block you
17:43 I'm actively using TTYDB already
18:30 We're going to tweak, improve and optimize the AI agents
18:40 The biggest issue right now
19:10 What's next for the channel?
20:00 We took this product from Zero to One
21:10 Here's what you can expect
21:35 Prelaunch Deal - 50% off for life
21:54 The series is over - I'm kinda sad - someones watching
22:49 One last thing
🐛 tags
#dataanalytics #promptengineering #aiprogramming
The upside of self-correcting agents in software engineering is pretty obvious. Save time, accomplish more while doing less, and worry less. It's a great step toward Agentic software engineering as a whole. Remember, we can't jump to step 10 and just 'build autonomous software'. We take small, sustainable steps the move us toward becoming true Agentic engineers. One experiment at a time. One application at a time. One product at a time. We'll also improve our Vite and Bun charged Vue.js front end with coding assistant AIDER running in the background while we tweak our Python/Flask backend. This is for software engineers and developers, this guide offers insights into enhancing software reliability and efficiency through AI-driven self-correction.
There's a lot of noise, drama and rapid news in the AI tech space right now and rightfully so. Focus going to be our most valuable tool in the age of AI.
Big shout out to everyone that's been staying focused, watching the channel, getting value, and taking their engineering and product building to the next level.
👍 THE CODEBASE
github.com/disler/multi-agent-postgres-data-analytics/tree/v9-self-correcting-assistant
💻 Talk to your database
talktoyourdatabase.com
🤖💻 AI Engineering Resources
LLM OS: youtube.com/watch?v=zjkBMFhNj_g
Assistants API: platform.openai.com/docs/assistants/how-it-works/objects
AI Pair Programming AIDER: https://aider.chat/
(1) Watch Part One
youtu.be/jmDMusirPKA
(2) Watch Part Two
youtu.be/JjVvYDPVrAQ
(3) Watch Part Three
youtu.be/4o8tymMQ5GM
(4) Watch Part Four
youtu.be/CKo-czvxFkY
(5) Watch Part Five
youtu.be/UA6IVMDPuC8
(6) Watch Part Six
youtu.be/XGCWyfA3rgQ
(7) Watch Part Seven
youtu.be/KwcrjP3vuy0
(8) Watch Part Eight
youtu.be/7EA19-D4-Zo
📘 Chapters
0:00 Diagnose, Generate, Execute
0:10 Tight Feedback Iteration Loops
0:20 LLMs can auto-correct any output system
0:33 Don't go all in on GPTs
0:53 Can LLMs run your software business?
1:30 Our Postgres Series is almost complete
1:50 Self Correcting Assistant - Recap
2:52 Firing off AIDER to work in parallel
4:25 Application Architecture - current and new
6:04 Channel 2024 Sneak Peek
6:25 Postgres Data Analytics Backend
7:22 An SQL query that will BOMB your database
8:21 New Self-Correcting Assistant Code
9:38 Turbo4 Assistants API wrapper class
14:35 Self-Correcting Assistant - RUN
16:11 Analyze the LLM corrected results
17:48 Agent Spyware Chat Conversation
18:20 Wow, it corrected a different error
19:25 All Great Devs have TIGHT feedback loops
20:45 Assistants API Pros, Cons, and Strategies
21:52 Build USE CASE specific AI Agents
22:35 The biggest WIN of the Assistants API
23:40 Assistants API is a heavy step forward
23:55 Negatives of Assistants API
25:47 Lot's of risk to our application
26:54 Should you go all in on Assistants API and GPTs?
28:00 A real world example of decided against Assistants API
28:42 Self-Correcting Assistants - Use Cases
29:15 The Channel has taken off but 2024 is going to be crazy
29:32 The End of the Series and what to expect
🐛 tags
#dataanalytics #agentic #promptengineering
We'll see firsthand how AI co-pilots, powered by the latest GPT-4 model, have become incredibly efficient and accurate. To showcase this, we're building the UI in three different frameworks – one of which I've never used before. This isn't just about coding; it's about AI Engineering our way to a minimal viable product for a data analytics frontend. Through 'Comment Copilot Coding', you write the comments, and AIDER fills in the code. It's a perfect blend of simple in editor, in context natual language prompting and AI efficiency. This video isn't just a tutorial; it's a glimpse into the future of software development, where natural language becomes the ultimate programming language. We'll open multiple terminals, run AIDER simultaneously, and utilize our multi-agent Postgres data analytics tool behind a simple python flask API to generate SQL and SQL results directly from our new UIs. Dive into the realm of multi-agent coding, demonstrating that with tools like AIDER and the expansive capabilities of GPT-4 Turbo, the real question shifts from: "what can I build?" to "How quickly can I build". Let's push the boundaries of AI in software engineering as we take one more step toward agentic engineering.
Aider Prompt: "Read the comments and wherever you see 'code:' generate the request. Keep all code generation in the component."
👍 THE CODEBASE
github.com/disler/multi-agent-postgres-data-analytics/tree/v8-ccc-ai-engineering-with-aider
🔥 TALK TO YOUR DATABASE
talktoyourdatabase.com
🤖💻 AI Engineering Resources
Aider: https://aider.chat/
Vite: https://vitejs.dev/guide/#scaffolding-your-first-vite-project
Bun: https://bun.sh/
Flask: flask.palletsprojects.com/en/3.0.x
Vercel Flask: vercel.com/new/templates/python/flask-hello-world
(1) Watch Part One
youtu.be/jmDMusirPKA
(2) Watch Part Two
youtu.be/JjVvYDPVrAQ
(3) Watch Part Three
youtu.be/4o8tymMQ5GM
(4) Watch Part Four
youtu.be/CKo-czvxFkY
(5) Watch Part Five
youtu.be/UA6IVMDPuC8
(6) Watch Part Six
youtu.be/XGCWyfA3rgQ
(7) Watch Part Seven
youtu.be/KwcrjP3vuy0
📘 Chapters
00:00 We're not typing this code
00:57 Vite, Bun, Aider
01:27 CCC
03:02 All Roads Lead To AIDER
04:27 Vue, Svelte, React
05:15 I've never used Svelte
06:07 It's not perfect but it's fast and close
06:31 Integrate our Postgres Data Analytics Tool
07:57 Hit the API without coding
10:20 GPT-4 Turbo, AIDER, Multi-Agent Results (LETS GO)
12:42 Kind of annoying but easily fixable
13:23 Svelte Reactivity Issue - One Prompt Fix
14:33 React Issue - Quick Fix - Sweet Results
16:17 Let's push this further - Local Storage Prompt
17:37 Vue, Svete, React - Local Storage Prompt
19:28 Use CCC to build applications faster
19:57 Add this to your toolbox
20:30 What's next?
21:01 Bro what about low code and no code?
21:56 TTYDB - Talk To Your Database - From Scratch
22:27 LLMs are changing the way we code
22:41 Stay on the edge of AI Engineering
🐛 tags
#promptengineering #aider #copilot
In this video, we delve into GPT-4 Turbo Assistants and their practical applications for builders and engineers. We'll explore a hands-on approach to integrating these technologies into our ongoing project: a Postgres data analytics tool that accelerates the way we interact with our SQL databases. Our focus is on practical, reusable Python structures, offering insights into how GPT-4 Turbo can enhance your product creation process with faster, safer, and more efficient solutions. We discuss the macro and micro implications of OpenAI's latest technologies and how we can stay on the edge of AI product engineering and enhance our personal engineering workflows.
We'll discuss the question of which technologies to focus on - AutoGen, Assistants API, or simple chat completion calls for your projects. Uncover the ideal solution tailored to your specific application needs, be it for automated tasks, context-rich interactions, or quick, straightforward queries.
This series isn't just theoretical – we're building real, applicable tools. In this video we create a reusable GPT-4 Turbo Assistant: TURBO4, and demonstrate its integration with our existing postgres multi-agent system. Our approach is clear and structured, aimed at maximizing the return on your investment in learning and applying these new technologies. We'll provide all necessary code links in the description. If you're intrigued by the potential of conversing with your database or eager to be among the first to test our upcoming 'talk to your database' product, the link will be in the description for that as well. Subscribe, like, and stay tuned for on the ground insights into utilizing the incredible power of GPT-4 Turbo for your engineering and product development endeavors.
👍 THE CODEBASE (Turbo4)
github.com/disler/multi-agent-postgres-data-analytics/tree/v7-turbo4-assistants-threads-messages
🔥 TALK TO YOUR DATABASE
talktoyourdatabase.com
🤖💻 AI Engineering Resources
OpenAI Assistance API: platform.openai.com/docs/assistants/overview
(1) Watch Part One
youtu.be/jmDMusirPKA
(2) Watch Part Two
youtu.be/JjVvYDPVrAQ
(3) Watch Part Three
youtu.be/4o8tymMQ5GM
(4) Watch Part Four
youtu.be/CKo-czvxFkY
(5) Watch Part Five
youtu.be/UA6IVMDPuC8
(6) Watch Part Six
youtu.be/XGCWyfA3rgQ
📘 Chapters
00:00 OpenAI Just Killed 1k+ Startups
01:35 What we can do on a daily basis
01:00 Recap Our Postgres Data Analytics Tool
05:05 Assistants API: Break it down
06:40 Turbo4 - Assistants API Wrapper Class
10:04 Turbo4 - API Deep Dive
17:14 Replace AutoGen Data Analytics Team with turbo4
20:56 Turbo4 Assistant In Action
24:00 Assistants API vs AutoGen vs Plain LLM
26:23 OpenAI Macro & Micro Strategy
27:30 Macro: OpenAI is becoming the Apple for LLMs
28:40 Micro: Building blocks, Composability, Reusability
29:50 The Ground Is Shifting Beneath Us
30:40 Shift and Move with the Ground
31:37 The Golden Nugget
32:00 Three Videos Left - 7 / 10
🐛 tags
#agentic #gpt #engineering