Opening Keynote, Michael Truell | Compile 26 Cursor 2026-06-22 | At this opening keynote of Cursor's inaugural Compile conference, co-founder and CEO Michael Truell traces Cursor's path to today. Michael, Kevin Niparko, and Tomas Reimers then announce Cursor Mobile, the Origin Git platform, and updates on a new model trained from scratch.0:00 - How Cursor started4:11 - Becoming agent-first5:10 - 95% of usage is now agents7:47 - Giving agents their own computer8:05 - Cursor as a platform9:55 - Building our own models13:18 - Updates to cloud agents18:02 - Cursor Mobile19:35 - Origin, an agent-native Git platform23:06 - A new model trained from scratch
Grok Bot For Product Best Practices Cursor 2026-09-03 | In this workshop, we cover how a product team of Bots research, write, design, and ship.Key workflows covered• Specialist teammates: You staff named Bots for chief of staff, analyst, PM, designer, and engineering manager. Each keeps its own memory and tools, so you DM the specialist for that job.• Attention list: A Bot watches Slack, Gmail, and Notion and lists where your attention went. Use that list to filter noise and compare focus with your stated priorities.• Routines: Ask a Bot to repeat a job on a schedule. We set hourly inbox triage on the chief of staff and an hourly install pulse after an X MCP launch.• Research to PRD: PM Pete pulled social, Slack, and user research into a Notion RFC for voice mode. Pixel used Figma and the design system to drop mocks into the same doc.• Ship with Cloud Agents: Pete and Emily joined a group chat. Emily split the PRD into tickets and handed coding to Cloud Agents, pausing when a login or approval was needed.Helpful resources• Grok Bot: https://x.ai/bot• Grok Bot for PMs: https://x.ai/bot/guides/grok-bot-for-pms• Grok Bot overview: docs.x.ai/grok-bot/overview• Upcoming workshops: cursor.com/workshops
Meet Grok Bot: Your Team of AI Agents Cursor 2026-09-03 | In this workshop, we introduce Grok Bot — a team of always-on agents with memory, tools, and their own computers — across personal and engineering use cases.Key features covered:Team of always-on agents:Grok Bot gives you persona-based bots that take on roles, delegate to each other, and message back and forth. You act as the CEO organizing your team.Three pillars:Every bot has persistent memory stored in S3 with no context limit, access to your connected plugins and accounts, and its own remote computer that keeps running after you close your laptop.Chief of Staff demo:Spin up inbox, calendar, and to-do bots that coordinate in a group chat. Organize your day by voice or text, and tag specific bots like you would in a group message.Skills & Teach a Task:Save shareable skills, or demonstrate a workflow on the bot's computer — like booking a flight — and it turns the demo into a reusable skill for gaps no plugin covers.Routines & controls:Trigger bots on a schedule, message, or incident. Set granular rules so a bot asks before sensitive actions, like emailing external vendors or editing a shared calendar.--------------------Helpful resources:Get Grok Bot: https://x.ai/bot--------------------This is a recording from our live session on August 20, 2026.For more events like this, check out upcoming workshops at cursor.com/workshops?source=attended_workshop
Grok Bot for GTM: From Prospecting to Customer Calls Cursor 2026-09-03 | In this workshop, Krista Letz (GTM at SpaceXAI) shows how she runs prospecting, call prep, slides, forecasting, and customer research on Grok Bot.Key workflows covered:Bots as specialized teammates:Each Grok Bot has a persona and a job, not one-off tasks. They work async — overnight, on calls, or away from your computer — and you onboard them like new hires.Chief of Staff & routines:Set routines to scan your inbox, flag and draft replies, and prep you for the day's calls. Ask Grok Bot what else you could automate, then turn repetitive work into routines.Prospecting bot:It researches accounts and prospects, watches their podcasts and webinars for personal hooks, pulls intent data, and drops messaging straight into your Gmail drafts at scale.Slides & engineer bots:The slides bot builds customer decks from Granola and Gong notes using your Figma brand kit. The engineer bot answers technical questions live from your codebase or docs.Forecast & customer experts:The forecast bot updates Salesforce next steps from email, Slack, Gong, and Granola. A per-customer expert tracks Slack, notes, feature requests, Linear tickets, and news.Skills, computer use & controls:Teach a skill by demonstrating a workflow, connect tools via plugins or the bot's own computer when no plugin exists, and set approval rules before sensitive actions.--------------------Helpful resources:Get Grok Bot: https://x.ai/bot--------------------This is a recording from our live session on August 26, 2026.For more events like this, check out upcoming workshops at cursor.com/workshops
Model Selection & Token Efficiency Cursor 2026-09-03 | In this workshop, we cover what tokens are, what burns them, and how to pick a model without slowing down.Key topics covered:Tokens:You are billed for tokens in and out of the model, not for agent actions like search. The harness wraps the model and is tuned per model.Four token types:Input, output, cache write, and cache read. Output costs more than input. Cache reads cost less, and a rule or model change can rewrite the cache.Context:The harness feeds context back each turn. Near 90% it compacts the middle of the chat, not the prefix or your latest ask.Model choice:Explore in Ask Mode, plan with a stronger model, then build with Composer. Auto can send harder turns to a stronger model.Habits:Start a new chat per task, mention past chats, and keep always-on rules short so they do not ride every turn.--------------------Helpful resources:Models and pricing: cursor.com/docs/models-and-pricingPlan Mode: cursor.com/docs/agent/plan-modeRules: cursor.com/docs/context/rulesSkills: cursor.com/docs/skills--------------------This is a recording from our live session on August 25, 2026.For more events like this, check out upcoming workshops at cursor.com/workshops
Refactoring Legacy Codebases Cursor 2026-09-03 | In this workshop, we cover a four-step way to migrate a legacy codebase with Cursor.Key workflows covered:Audit with Canvas:Use /canvas to map test coverage and which components to migrate. The demo audits a WordPress repo, then shares the canvas as a living document.Plan Mode:Write the migration as a markdown strategy with requirements and steps. Plan Mode does not write code, so you can review the approach first.Tickets from a plugin:Use a plugin, like Atlassian, to split that plan into Jira tickets. Plugins can also bring the vendor's skills with the MCP.Cloud agents:Hand those tickets to a cloud agent that keeps working with your laptop closed. It can use the same MCPs and come back with pull requests, screenshots, and videos.Automations:Schedule a cloud agent, like a weekly feature-flag cleaner, so the migration does not rot. You can also start from templates for coverage, scans, or incident triage.--------------------Helpful resources:Plan Mode: cursor.com/docs/agent/plan-modeCloud Agents: cursor.com/docs/cloud-agentAutomations: cursor.com/docs/automationsPlugins: cursor.com/docs/plugins--------------------This is a recording from our live session on August 13, 2026.For more events like this, check out upcoming workshops at cursor.com/workshops
Nokia analyzes 50M+ lines of code in two weeks with Cursor Cursor 2026-09-02 | Kal De, SVP of Product & Engineering at Nokia, shares how his team is redesigning their entire SDLC with Cursor. In just two weeks, Nokia used Cursor to analyze more than 50 million lines of code and develop an evidence-based plan to transition a monolithic network function toward a more distributed, service-based architecture.Nokia is also using Cursor to turn complex, coordination-heavy work into streamlined engineering workflows. In less than a week, one engineer went from an initial problem statement to a detailed PRD and a working project-management tool – automating and visualizing roughly 80% of a process that previously required coordination across 6–10 project and program managers.When customer-impacting defects arise, Cursor helps Nokia cut root cause analysis from weeks to days, accelerating the path to a patch.These results point to a broader shift in how Cursor is transforming Nokia’s approach to software engineering.Learn more about how Cursor helps teams transform the full software development lifecycle: cursor.com/product
Benchling cuts migration time by 75% with cloud agents Cursor 2026-07-29 | Emily Reed, Developer Productivity Software Engineer at Benchling, shares how her team used Cursor to lift a complex local development environment into the cloud, cut down manual setup, and take on a large-scale test migration that had been slowing a major React upgrade.Instead of asking teams to migrate thousands of legacy Enzyme tests one by one, Benchling orchestrated hundreds of agents in parallel and brought the timeline down from one year to one quarter. Today, about 30% of Benchling’s merged pull requests come from cloud agents. After rolling them out, Benchling saw a 20% jump in developer satisfaction, and 95% of developers now say they’re satisfied with their AI tooling.Learn more about cloud agents: cursor.com/docs/cloud-agent
Introducing Cursor for iOS Cursor 2026-06-29 | Build from anywhere by launching always-on cloud agents. Or remotely control agents running on your computer from the app.Learn more: cursor.com/blog/ios-mobile-app
Notational Intelligence, Linus Lee | Compile 26 Cursor 2026-06-29 | Linus Lee, Engineer at Thrive Capital, on notational intelligence: how the ways we write ideas down shape our thinking, and how deep learning might invent entirely new notations.00:00 Notational intelligence01:27 What makes good notation02:28 Abstraction03:18 Suggestiveness and natural transformations05:28 Graphical notation06:27 The coordinate plane and the arrow08:14 Programming languages as notation09:28 Inventing new notation with deep learning11:13 Building the toy model13:42 The handout and training results15:03 Invariants that make symbols meaningful15:56 Our world vs. an alien world of ideas16:30 Models as a simulator for anything
Agency in Language, Alane Suhr | Compile 26 Cursor 2026-06-29 | Alane Suhr, Assistant Professor at UC Berkeley, on agency in language: the kinds of meaning we make with words, what a language model actually is, and why we, not the concept of AI, hold the agency.00:00 Agency in language06:31 The reference of AI today06:57 What a large language model is09:10 Instruction tuning and RLHF10:20 The connotation of AI13:13 We have agency, not the concept
Intelligence Efficiency, Ben Geist | Compile 26 Cursor 2026-06-29 | Ben Geist, Research Engineer at Ramp, on intelligence efficiency: why paying more for tokens yields diminishing returns, and how better context, not just more compute, makes models more efficient.00:00 Intelligence efficiency01:00 Smarter models aren't more efficiently smart02:04 Token spend data from Ramp03:01 Entropy reduction machines03:45 Two views of entropy05:01 Context, not just work06:09 Three examples in latent space06:39 Multi-agent systems with shared context08:56 In-context learning and sparse attention
The Memory Problem, Baseten | Compile 26 Cursor 2026-06-29 | Mudith Jayasekara, Charlie O'Neill, and Harry Partridge of Baseten's research team on the memory problem for long-horizon agents: compressing the KV cache to get near-lossless retrieval of relevant context at inference time.00:00 The memory problem03:15 Selection vs. synthesis04:22 Amortization and sparse autoencoders05:38 Compressing the KV cache08:39 Iterative compaction10:41 A compacted cache is an MLP
Explaining Culture to Technology, Paul Ford | Compile 26 Cursor 2026-06-29 | Paul Ford, Co-founder of Aboard, on what technologists can learn from how magazines actually work, and why culture is a distributed, lossy prediction model rather than its output.00:00 Explaining culture to technology00:58 How a magazine actually works03:09 Rhetoric greater than facts04:03 What culture is05:27 A model of consciousness06:15 Culture as an operating system07:17 The tech industry and risk
The New PM, Claire Vo | Compile 26 Cursor 2026-06-24 | Claire Vo, founder of ChatPRD, on what product management becomes when code is abundant but the market is not.0:00 - Anybody can build anything1:38 - Inventing ways to not build7:13 - The constraint has shifted9:42 - The age of abundance11:51 - What product should be now13:03 - Build something people want15:41 - Manifesting novel ideas17:50 - Bringing people along20:17 - Is PM dead?24:42 - Find the money, build companies
What Is Your Job Now, Farhan Thawar | Compile 26 Cursor 2026-06-24 | Farhan Thawar, head of engineering at Shopify, on what an engineer's job becomes when AI writes most of the code.0:00 - What is your job now?0:55 - How the SDLC is shifting3:13 - Learning is the collateral4:33 - The bottleneck always moves7:58 - Tobi's memo11:17 - Hiring 1,000 interns12:48 - The end of the centaur19:39 - Prototype is not production20:46 - How Shopify works with AI25:22 - What changed, what stayed
Agents and Infrastructure, Sam Lambert | Compile 26 Cursor 2026-06-24 | Sam Lambert, CEO of PlanetScale, on how infrastructure must change for agents, with a live demo of agents optimizing and sharding a real database.0:00 - Infrastructure for agents1:18 - A talk run by agents3:10 - Agents optimize the database6:00 - Branching, deploys, and safety8:36 - Catching and undoing a bad change11:23 - Why sharding is hard14:02 - Refactoring for sharding19:28 - Built for humans first21:34 - A safe loop for agents24:56 - Infrastructure, not just smarter models
Closer to the Material, Ryo Lu | Compile 26 Cursor 2026-06-23 | Ryo Lu on how AI changes the way we build and what it must not erase.0:00 - Building ryOS1:21 - What should exist?2:11 - The loop AI changes4:17 - The black box risk6:43 - Output vs. material8:31 - The Glass interface11:14 - Prototyping Glass with Cursor14:03 - When software felt alive17:36 - Where craft moves20:25 - A more human future
Running 128 Coding Agents at Once Cursor 2026-06-11 | Cursor's Sam Whitmore sits down with Baseten's Charlie O'Neill and Harry Partridge to discuss where working with agents goes next: running them by the hundred, getting them to message and review each other, and building systems instead of just managing parallel tasks.
How are coding agents changing software engineering? Cursor 2026-05-29 | New data from cursor.com/insights.
Training Composer 2 Cursor 2026-05-21 | In this workshop, Sasha Rush walks through how Cursor's research team builds Composer 2 - from base model choice to long-horizon reinforcement learning.Key topics covered:Base model selection:Composer 2 starts from Kimi K2.5 (1T params, 32B active, 256K context). The choice came down to both infrastructure fit and initial benchmark scores.Continued pre-training:A coding-focused pre-training stage builds domain knowledge. More tokens here translate into measurably higher rewards after the RL stage.Long-horizon RL with auto-install:Composer 1.5 bootstraps each training environment by exploring the repo, generating install commands, and writing verification tests before RL begins.Reward shaping and self-summarization:A nonlinear length penalty balances speed and depth, while self-summarization lets the model continue past its context limit and still share one final reward across the rollout.Cursor Bench:An internal eval of short, ambiguous prompts and large multi-file diffs from real engineer queries. It separates strong from weaker models much more cleanly than SWE-bench.-----------------------Helpful resources:Models: cursor.com/docs/modelsCursor docs: cursor.com/docs-----------------------This is a recording from our live session on May 14, 2026.For more events like this, check out upcoming workshops at cursor.com/workshops
Simon Eskildsen on scaling Shopify, building turbopuffer, and the future of databases Cursor 2026-05-14 | Sualeh Asif (Co-Founder of Cursor) sits down with Simon Eskildsen (CEO of turbopuffer) to discuss scaling Shopify through flash sales, why a new database company emerges every 15 years, and what makes a P99 engineer.0:00 - Scaling Shopify through flash sales and outages7:11 - How the Shopify infrastructure team evolved8:13 - How top infrastructure teams collaborated in the 2010s10:35 - Engineering principles from Logrus and on-call17:38 - The story behind Simon’s famous-ish blog, Napkin Math23:05 - Why new database companies keep winning32:21 - How Simon became a fan of databases35:45 - AI coding, and where agents still fail38:44 - What it would take for turbopuffer to become Google-scale42:10 - Hiring P99 engineers in the AI era48:45 - What's next for databases
How Intuit, DoorDash, and Atlassian are adopting AI coding Cursor 2026-05-12 | Jordan Topoleski (COO, Cursor) moderates a panel with engineering leaders from Intuit (Chris Kasten), DoorDash (Ryan Sokol), and Atlassian (Taroon Mandhana) on adopting AI coding tools at scale. They cover what's actually changed in their development workflows, how they measure productivity gains, and what advice they'd give other engineering leaders making this transition.0:00 - How AI coding took hold at Intuit, DoorDash, and Atlassian3:41 - What's changed most across the SDLC6:18 - Re-platforming three companies with AI at DoorDash7:47 - The bell curve is inverting as AI eats the middle of the SDLC8:14 - Atlassian's engineers spend 15% of time coding and 85% on everything else11:17 - Measuring the actual impact of AI coding tools13:42 - Small pods delivering nine-month roadmaps in a month at DoorDash15:15 - Intuit's alpha teams and the proof-to-release package loop17:17 - One piece of advice for engineering leaders adopting AI
What happens when agents get their own computers Cursor 2026-05-12 | Jonas Nelle (Engineering Lead, Cursor) demos two new capabilities that remove human bottlenecks between building and shipping: agents that test their own code and produce video proof, and a cloud agent tha babysits your PR to merge.0:00 - Replacing human bottlenecks with intentional checkpoints0:51 - Agents that write code, test it, and produce video proof1:48 - Reviewing a feature by watching the agent's demo video2:02 - Taking direct control of the agent's cloud environment3:12 - The new Cursor 3 agents-first view3:53 - From PR created to merge-ready without human busywork5:04 - Building your own agent systems with Cursor's primitives
How Cursor builds agentic workflows across the SDLC Cursor 2026-05-12 | Tido Carriero (VP of Engineering, Cursor) shares how the team is building agent systems across the entire SDLC—from bug triage bots to security reviewers to growth experiment pipelines—and what it takes to break past the 40% productivity plateau.0:00 - Most companies plateau at 40% more productive with AI coding assistants2:39 - The broader SDLC is the real bottleneck, not just writing code3:03 - Figuring out which steps still need humans5:12 - The threshold problem and removing the rest of the bottlenecks5:34 - Turning intelligence into software with polished agent artifacts6:53 - Building agent teams outside the core build phase7:20 - The PM agent and EM agent that started from a Slack channel10:48 - The architecture of a fully autonomous bug triage system12:59 - A security bot that has fixed over two hundred vulnerabilities14:22 - Agentic risk detection that lets low-risk PRs skip human review15:52 - Four agents that transformed growth experimentation throughput18:53 - Organizational shifts for the agent era
The next era of AI coding Cursor 2026-05-12 | Michael Truell (Co-Founder and CEO, Cursor) walks through the evolution of software engineering, from writing every line by hand to managing teams of autonomous agents, and shares data on how fast the shift is happening.0:00 - The Star Wars premiere and its place in technology history1:34 - The tedium of building software in formal programming languages2:58 - Why software complexity is hidden and what that costs4:22 - The explosion of agent requests vs. tab accepts in 20254:57 - 30% of Cursor's PRs are fully agent-developed end-to-end5:29 - Enterprise code went from 15% to 75% AI-generated in a year5:53 - Engineers are becoming agent managers7:36 - Working with dozens of parallel agent colleagues8:01 - Agents building a browser in a week with no humans in the loop
Beyond efficiency: PayPal expands whats possible to build with AI Cursor 2026-05-11 | PayPal could deliver 40% more capabilities in 2026. High-adoption teams are now deploying daily and completed a 3,000-app Java upgrade 6x fasterLearn more: cursor.com/blog/paypal
Introducing Cursor 3 Cursor 2026-04-02 | We’re introducing Cursor 3. It is simpler, more powerful, and built for a world where all code is written by agents, while keeping the depth of a development environment.Learn more here: cursor.com/blog/cursor-3
Introducing Automations: always-on coding agents Cursor 2026-03-05 | We're introducing Cursor Automations for building always-on agents.These agents run on schedules or are triggered by events like a sent Slack message, a newly created Linear issue, a merged GitHub PR, or a PagerDuty incident.You can learn more here: cursor.com/blog/automations
Software is changing Cursor 2026-02-26 | Micromanagement isn't the future of software engineering.cursor.com/onboard
A computer for every agent Cursor 2026-02-24 | Cursor can now onboard to your codebase, run in cloud sandboxes, and send you video demos of their work (rather than just diffs).cursor.com/onboard
Cursor now shows you demos, not diffs Cursor 2026-02-24 | Agents in Cursor can use the software they build and send you videos of their work.Try it at cursor.com/onboard.Read our announcement: cursor.com/blog/agent-computer-use.
Box chooses Cursor for enterprise-grade quality, security, and control Cursor 2026-02-13 | Over 85% of developers at Box now use Cursor daily, driving a 30-50% increase in roadmap throughput and 80-90% reduction in effort for codebase migrations. Learn more here: cursor.com/blog/box
Dropbox uses Cursor to index over 550,000 files and build an AI-native SDLC Cursor 2026-01-26 | Dropbox accepts more than 1 million lines of agent generated code with Cursor every month, improving PR velocity and cycle time. Read more here: cursor.com/blog/dropbox
Salesforce ships higher-quality code across 20,000 developers with Cursor Cursor 2026-01-23 | Over 90% of developers at Salesforce now use Cursor, driving double-digit improvements in cycle time, PR velocity, and code quality.Learn more here: cursor.com/blog/salesforce
John Schulman on dead ends, scaling RL, and building research institutions Cursor 2025-12-17 | A conversation with John Schulman on the first year LLMs could have been useful, building research teams, and where RL goes from here.00:00 - Speedrunning ChatGPT09:22 - Archetypes of research managers11:56 - Was OpenAI inspired by Bell Labs?16:54 - The absence of value functions18:23 - Continual learning21:09 - Brittle generalization24:05 - Co-training generators and verifiers, GANs27:06 - John’s personal use of AI for research28:54 - Day in the life33:01 - Slowdowns in consequential ML ideas36:21 - "Peer review" within the labs39:19 - Distribution shift in researchers43:33 - Future of RL45:33 - Will the labs coordinate if the world needs them to?44:46 - Forecasting ills in AGI and engineering47:53 - Thinking Machines
A new visual editor: design directly in your codebase Cursor 2025-12-11 | You can now design directly in your codebase.Select elements, modify them visually, and Cursor writes the code.cursor.com/blog/browser-visual-editor
Coding agents for the hardest bugs Cursor 2025-12-10 | Cursor can now fix your trickiest bugs.Debug Mode instruments your code, spins up a server to capture logs, and streams runtime data to the agent.
Introducing Cursor 2.0 Cursor 2025-10-29 | Introducing Cursor 2.0.Our first coding model and the best way to code with agents.
Cursor Agent: 10 Pro Tips! Cursor 2025-10-11 | 0:00 – 1. Plan mode2:53 – 2. Context menu4:17 – 3. Custom commands5:56 – 4. Images7:55 – 5. Duplicating chats8:19 – 6. Context visibility8:51 – 7. Usage visibility9:15 – 8. Keyboard shortcuts9:37 – 9. New chats10:12 – 10. Checkpoints11:20 – Bonus round
Introducing Plan Mode Cursor 2025-10-07 | Cursor can now write detailed plans before starting complex tasks.This allows agents to run for significantly longer.Plans are generated as Markdown and can be edited inline or stored in code. Enter plan mode with Shift+Tab in the agent input.cursor.com/blog/plan-mode
AI Foundations: How AI Models Work Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
AI Foundations: Tokens & Pricing Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
AI Foundations: Introduction Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
AI Foundations: Hallucinations Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
AI Foundations: Context Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
AI Foundations: Tool Calling Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
AI Foundations: Agents Cursor 2025-09-27 | Read all chapters, check your knowledge, and try AI models at cursor.com/learn.
Using the Figma MCP server with Cursor Cursor 2025-09-26 | Learn how to convert your Figma designs into code with Cursor and the new Figma MCP.View and install MCP servers: cursor.com/docs/context/mcp/directory
Our designer built an operating system with Cursor Cursor 2025-07-30 | Ryo Lu, Cursor's Head of Design, on his personal operating system, ryOS, and what it means to be a builder with AI.Play with ryOS:https://os.ryo.lu
