Uploaded August 2026 | Updated September 2026, 2 weeks ago
Check Out Yorby, the social media marketing tool for startups: https://www.yorby.ai?utm_source=yatb-yt
How I built a better AI agent after my first version completely failed.
I spent the past month building an AI agent inside my startup, and the first version was honestly pretty bad. In this video, I break down what went wrong, what I changed, and how I rebuilt the agent to actually produce useful results.
I cover how I think about AI agent architecture, tool calling, planning modes, clarifying questions, context gathering, and why the quality of your input matters more than the agent framework you choose. I also walk through the real AI carousel agent I built for Yorby, my AI social media marketing app, and show how it plans, asks questions, gathers context, and creates better output for users.
If you are building AI agents, working with the Vercel AI SDK, adding tool calling to your app, or trying to build better AI products in 2026, this is the practical breakdown of what actually worked for me.
Chapters:
00:00 I Built an AI Agent for My Startup
00:54 Why Agent Frameworks Don’t Matter
01:48 The Biggest Mistake I Made
02:42 The AI Agent I Built for Yorbi
04:03 Why One-Shot Prompting Failed
04:57 Plan Mode and Context Gathering
05:08 Asking Clarifying Questions
05:55 How the Agent Reviews, Plans, and Iterates
07:01 Why Input Quality Matters Most
07:21 General Purpose Tools vs Overbuilt Tools
07:58 Copying What Works From AI Coding Tools
08:40 How to Build Better AI Agents in 2026
Want to work with me 1:1? Book some time with me at youraveragetechbro.com
Check out my AI-powered interview prepping tool: perfectinterview.ai/?utm_source=yatb-yt
Check out my latest SaaS product to start automating your job: montee.ai/?utm_source=yatb-yt
Follow me on TikTok: tiktok.com/@youraveragetechbro
Follow me on Instagram: instagram.com/youraveragetechbro
#AIAgents #AICoding #BuildInPublic
Check Out Yorby, the social media marketing tool for startups: https://www.yorby.ai?utm_source=yatb-yt
How I built a better AI agent after my first version completely failed.
I spent the past month building an AI agent inside my startup, and the first version was honestly pretty bad. In this video, I break down what went wrong, what I changed, and how I rebuilt the agent to actually produce useful results.
I cover how I think about AI agent architecture, tool calling, planning modes, clarifying questions, context gathering, and why the quality of your input matters more than the agent framework you choose. I also walk through the real AI carousel agent I built for Yorby, my AI social media marketing app, and show how it plans, asks questions, gathers context, and creates better output for users.
If you are building AI agents, working with the Vercel AI SDK, adding tool calling to your app, or trying to build better AI products in 2026, this is the practical breakdown of what actually worked for me.
Chapters:
00:00 I Built an AI Agent for My Startup
00:54 Why Agent Frameworks Don’t Matter
01:48 The Biggest Mistake I Made
02:42 The AI Agent I Built for Yorbi
04:03 Why One-Shot Prompting Failed
04:57 Plan Mode and Context Gathering
05:08 Asking Clarifying Questions
05:55 How the Agent Reviews, Plans, and Iterates
07:01 Why Input Quality Matters Most
07:21 General Purpose Tools vs Overbuilt Tools
07:58 Copying What Works From AI Coding Tools
08:40 How to Build Better AI Agents in 2026
Want to work with me 1:1? Book some time with me at youraveragetechbro.com
Check out my AI-powered interview prepping tool: perfectinterview.ai/?utm_source=yatb-yt
Check out my latest SaaS product to start automating your job: montee.ai/?utm_source=yatb-yt
Follow me on TikTok: tiktok.com/@youraveragetechbro
Follow me on Instagram: instagram.com/youraveragetechbro
#AIAgents #AICoding #BuildInPublic
