Skip to content

About Dosu

A story about knowledge, code, and what comes next.

In 1969 Margaret Hamilton stood next to the Apollo guidance software all printed out and stacked taller than she is. That was how knowledge was captured and stored back then. Every decision and every calculation her team thought through was written down and physically saved.

Since then, we’ve gone from printed manuals to wikis. Then to Confluence, Notion, Slack threads, Jira tickets, design docs, post-mortems, and onboarding guides. The tools multiplied, but the core problem never changed: knowledge lives in people’s heads, and every system we’ve built to get it out depends on those same people finding time to write it down and keep it current.

But it rarely gets written down properly. Knowledge goes stale or gets lost entirely, and we’re forced to relearn what someone already knew.

Margaret Hamilton beside the Apollo guidance software source code, printed and stacked taller than she is

AI agents writing code have pushed those systems past their breaking point.

The trouble with agents is that they don’t have instincts. They trust whatever context they’re given and execute at machine speed. When the environment around them is hard to work in, they don’t push through the way an experienced engineer would. They stall, guess, or stop and ask a human for help.

And every new session starts from scratch, rediscovering the same complexity the last one already fought through.

That’s why we built Dosu.

Dosu is self-improving developer experience for coding agents. It learns the recurring sources of friction in your agents’ sessions, sends them runbooks so they keep moving, and finds root-cause fixes for the processes that keep tripping them up. It doesn’t just remember what an agent did last time. It shows the agent the best way through, then makes that way easier.

We believe developer experience for agents will matter as much as developer experience for people ever did. Agents do their best work when they can verify it with tests and benchmarks, in an environment that doesn’t need a human to unblock them. The teams that build that environment will get more done with less steering, and every session will make the next one smoother.

We saw this problem first because we lived it, as open source maintainers and engineers at fast-moving startups where knowledge was constantly being lost and relearned.

And today, over 200,000 engineers use Dosu. We’re SOC 2 certified. And we’re still a small, ambitious team building from San Francisco.

The organizations that thrive in the age of AI won’t be the ones with the best models. They’ll be the ones whose agents can do the work without a human steering every step.

We’re building what gets them there.

A robot agent works a control lever inside a brass frame while Dosu tightens a bolt on the frame beside it.

This is the way.

If you believe in an AI-native future and want to help agents do their best work, we’d love for you to be part of what we’re building.

Devin Stein's signature

- Devin Stein, CEO & Founder

Get your agents unstuck

Sign up for early access and we’ll help you get started with Dosu.