Deep dive with Naman, co-founder of Manicule — a YC Spring 2026 studio that writes end-to-end documentation for developer tools companies, with a specific focus on making those docs readable by AI agents, not just humans.

Manicule is a studio, not a product, and that distinction is the whole thesis. They work with seed to Series B developer tools companies that know documentation matters but do not have a dedicated DevRel function or the founder bandwidth to build it. Manicule comes in for the full process: talk to the engineers, understand the codebase, design the information architecture, write the pages, verify the code, create diagrams and screen recordings, and then maintain the docs so they do not drift out of sync with the product. They ship a complete project in two weeks, which Naman says is the fastest in the industry against a baseline closer to a month for a single writer.

The technical core is a pipeline of internal sub-agents. A client installs a GitHub app, and Manicule’s agents explore the codebase, build a separate repository of context about how the product works, and carry that understanding into the writing process. The same approach works for UI-based products, where agents navigate the actual interface to understand it rather than just reading code.

The part that fits this show is Manicule’s bet on documentation for agents. Naman’s view is that a human reading a page to learn something needs content written in a fundamentally different style than an agent consuming the same page. One concrete finding: adding an explicit anti-pattern section to agent-facing content measurably improves how well agents use a product, because some mistakes are so ingrained that an agent will keep repeating them unless told directly not to. Manicule writes separate content for humans and agents on the same docs, and is benchmarking approaches as the field figures itself out in real time.

Naman is direct about why this is not a fully automated product. Agents can produce a first draft, but he still spends around two hours per page bringing it to a publishable standard. His argument tracks the Sequoia thesis that AI-native services are the next big category, because clients do not want docs, they want outcomes, and outcomes require accountability that a standalone tool cannot provide.

Also covered: how Manicule got into YC by pitching a completely different idea and pivoting in the interview room, why the founders first applied to YC at 15, why a billion-dollar competitor in your space is not the threat founders think it is, and a fully manual cold email strategy that caps at five or six sends a day.

—

Agentic Stories is the weekday briefing on the AI agent economy — governance, security, and deployment. Deep Dives drop on off-days with founders building in the space. New episodes Monday, Wednesday, Friday.

agenticstories.ai