ContextWindow is a weekly publication about AI product management — the strategy, the technical depth, and the career path. Not a surface-level overview. The real thing.
Who writes this
I’m Michael Kent. I build AI product systems in production — RAG pipelines, MCP servers, agent frameworks — and manage the product strategy around them. I’m based in the Bay Area and actively transitioning into AI PM roles at high-growth companies.
I also run PostSynaptic, a technical build log where I show the actual code behind the concepts I write about here.
What this channel covers
- Technical: RAG architecture, eval design, MCP & agents, LLM tradeoffs
- Career: AI PM job search, portfolio building, salary signals, interview prep
- Craft: Prompting patterns, PRDs for AI features, roadmapping AI, metrics & evals
What this channel is not
It’s not a beginner channel. If you need someone to explain what an API is, there are better places to start. Come back when you’re ready to go deeper.
It’s not a “best practices” channel in the corporate sense. I share what works in practice, what I’d do differently, and where I cut corners and why. That’s more useful than a sanitized version of how things should work in theory.
The name
Everything that happens in an LLM interaction is shaped by what’s in the context window. The model has no memory, no knowledge, no judgment outside of it. Its behavior is entirely a function of what you put in that window.
Most PMs are working with a limited context window when it comes to AI. This channel is about expanding it.