Develop Yourself Build This AI Project and You’re Ahead of 90% of Developers
Mar 30, 2026
A hands-on walkthrough for building an AI advisory board that answers questions in the style of real influencers. Steps include calling an LLM programmatically, adding system prompts and guardrails, and creating a simple backend and frontend. Learn how to fetch and organize YouTube transcripts, inject relevant data with naive RAG, and know when to add vector databases.
AI Snips
Chapters
Transcript
Episode notes
Building With AI Vs Using AI Tools
- Building with AI (using APIs and grounding with data) is much rarer than using AI tools and gives a large advantage.
- Brian Ginny estimates only ~10–15% of developers build AI into products, so doing this puts you ahead of ~90–95% of peers.
Tried To Hire An AI Engineer And Couldn't
- Brian Ginny recounts trying to hire an AI engineer and failing, so he trained an existing team member instead.
- He used this example to show how rare AI engineering skills were and why salaries for AI engineers are rising fast.
Start By Calling An LLM Programmatically
- Get an API key and call an LLM programmatically as the first step to building AI features.
- Brian Ginny recommends Google AI Studio/Gemini for free testing, then add the key to .env and wire a backend route to forward user messages to the model.