“The workflows I described are real, based on actual commands and prompts I’m using in production code. But I have abstracted them a bit to make them easier to understand in the context of an arbitrary agent, not tied to specific idiosyncrasies of the tool I happen to be using at the moment”
awesome stuff. this was a great read. appreciate the realistic base here. definitely a message worth pushing right now.
The layer analysis is the right frame. The one I'd add: tools without an architectural memory layer push the context-management cost onto the user. Every session starts from zero, so the human is the integration layer. Agents that persist architectural decisions across sessions (a PRD, not just a chat log) are the ones that stop feeling like coding with amnesia.
“The workflows I described are real, based on actual commands and prompts I’m using in production code. But I have abstracted them a bit to make them easier to understand in the context of an arbitrary agent, not tied to specific idiosyncrasies of the tool I happen to be using at the moment”
awesome stuff. this was a great read. appreciate the realistic base here. definitely a message worth pushing right now.
Thanks! I'll follow up with lots of details.
awesome! honestly i’d love that. always good to hear from people actually building and working in the space.
The layer analysis is the right frame. The one I'd add: tools without an architectural memory layer push the context-management cost onto the user. Every session starts from zero, so the human is the integration layer. Agents that persist architectural decisions across sessions (a PRD, not just a chat log) are the ones that stop feeling like coding with amnesia.