Thank you! I have been dying for exactly this explanation before fully diving in. I'm ready, I built a server on a raspberry pi 5 and installed stuff and just stared at this thing and it works and I built it but don't know wtf to do next and won't start until someone explains in a way I can understand. I understand exactly what you are saying and that reinforces that I can do it I just need help from a source I trust who can help is change the world for good.
Great Pt 1 super interesting topic. The stage framework adapted from Yegge is really useful, and the "clarity of intent over technical skill" frame is great.
One observation / question I have: is the Stage 2 to Stage 3 line (read-only conversational AI vs. write-enabled agentic AI) not less firm? For example, Stage 2 tools can pick up Stage 3 capabilities via MCP. Claude.ai (to stick with Yegge :) with the Obsidian MCP, for example, gets full read and write access to a markdown vault from inside a conversational interface, (controlled, since agents can be more curious than is good for them) but there would be no Cursor or VS Code overhead.
When will part 2 come out? Very excited for this series.
I appreciate you teaching style. Do you have anticipated release date for this course?
Just came back to read this again, and very excited for this course!!!
Thank you! I have been dying for exactly this explanation before fully diving in. I'm ready, I built a server on a raspberry pi 5 and installed stuff and just stared at this thing and it works and I built it but don't know wtf to do next and won't start until someone explains in a way I can understand. I understand exactly what you are saying and that reinforces that I can do it I just need help from a source I trust who can help is change the world for good.
Thank you
Casey, I am looking forward to part 2. Currently applying to PhD in Mechanical Engineering.
Casey excited for part 2.
This is great! I’m so excited for part 2
Great Pt 1 super interesting topic. The stage framework adapted from Yegge is really useful, and the "clarity of intent over technical skill" frame is great.
One observation / question I have: is the Stage 2 to Stage 3 line (read-only conversational AI vs. write-enabled agentic AI) not less firm? For example, Stage 2 tools can pick up Stage 3 capabilities via MCP. Claude.ai (to stick with Yegge :) with the Obsidian MCP, for example, gets full read and write access to a markdown vault from inside a conversational interface, (controlled, since agents can be more curious than is good for them) but there would be no Cursor or VS Code overhead.
Looking forward to Pt 2!
The real question is what AI did you use to assist you publish/post this article?😂
Great read Casey
I’m excited for this series!