As an engineer, I’ve worked everywhere from startups to public companies, and one thing has always been true: documentation is rarely up-to-date. Without reliable docs, when people run into issues or have questions there are only two options: dig through code or ask for help. Most people choose the latter.

And, when people ask for help, everyone typically pings the same “go-to” engineer, or at the companies I’ve worked at, they ping me 🥲.

As much as I love unblocking my co-workers, it began to take up over 50% of my time. This experience put me on the path to build Dosu and is why I believe that AI may finally bring us to the promised land of complete, always up-to-date documentation.

When Teams Ship Fast, Docs are the First to Fall#

If you’ve ever worked on a serious software project, you know the drill. You ship code, you answer questions, you triage issues… and somewhere in between, you’re supposed to update the docs. Except, you never quite get around to it. Or, if you do, it feels like you’re the only one doing it.

The problem isn’t just that updating docs is boring (though, let’s be honest, it often is). It’s that it’s time-consuming, and now, with AI coding tools the rate of knowledge change is only accelerating. New features, bug fixes, and workarounds pile up faster than anyone can document them. As projects grow—with more contributors, more users, and more stakeholders—it becomes impossible for any one person to keep everything up to date.

This happens all the time. Take this message to me the other day from a user of a project I maintain about a new feature:

A contributor added a new feature, but they, me, and the maintainers forgot to document it–despite best efforts.

This happens internally at companies too. Consider a sales person who needs to find out, on behalf of a potential customer, whether the product offering generates content in markdown or something else. If the documentation is missing (it likely is), they’ll either have to track down an engineer or figure it out on their own—both of which slow down the team, introduce risk, and aren't a good use of anyone's time.

Maintaining Docs is an AI Job#

Like a calculator, AI doesn’t get tired, forgetful, or bored. It doesn’t procrastinate or deprioritize documentation. AI can monitor code changes, conversations, and support channels in real time, making sure the latest information is always reflected in the docs. When a new answer is shared or a bug is fixed, AI can automatically update the documentation or flag inconsistencies—something humans often miss or delay.

AI excels at repetitive, labor intensive tasks and maintaining documentation is just that. In order to keep documentation complete and up-to-date, you need to review every code change, ticket, and conversations for

1. Find relevant existing documentation
2. If there is related information, check to see if the new information is consistent. Does it support the existing information or contradict it
3. If there is no relevant information, should we start a new document for it?
4. Make changes
5. Repeat

Humans (try to) do this implicitly as they work, but AI can do this explicitly, in real time.

Making ‘Living Documentation’ a Reality#

People have dreamed of living documentation–documentation that evolves as fast as your code–for decades but human bandwidth to review and update documentation has always been the bottleneck.

Previously the only way to truly achieve living documentation was to treat code as your only documentation, but the reality is that most people–even engineers–have a hard time understanding code.

Now with AI, we have a way to couple written knowledge to code and get the best of both worlds: easy to understand documentation that evolves with your code.

At Dosu, we are making this a reality. Dosu is a living knowledge base–backed by code–that every team can understand—from engineering to operations.

Dosu not only maintains knowledge, but lets you interact with wherever you work. You can have Dosu answer questions and update docs from Slack and preview documentation changes on pull requests.

So if you’re tired of outdated docs, wasted time, and frustrated users, take a look at what purpose built and well considered AI systems like Dosu can do. I’m betting on it—and I think you should too.

P.S. Want to see it in action? Check out Dosu—we’re already helping thousands of teams save time and we’re just getting started. If you’d like to discuss this article, give general feedback or ask questions, feel free to join our Discord.