Promptless Fixes the Docs Nobody Wants to Update

The Macro: Documentation Is Always Wrong

Every software company has the same problem. The product ships faster than the docs can keep up. A feature changes in a Tuesday deploy. The API reference still describes the old behavior. A new integration goes live. The how-to guide doesn’t mention it for three weeks. A support ticket reveals a common confusion. Nobody updates the troubleshooting page because the engineer who could write it is already working on the next sprint.

This isn’t a discipline problem. It’s a structural one. Documentation is a second-class citizen in the development workflow. Engineers write code, ship PRs, and move on. The docs update is a separate task that lives in a separate system, often owned by a different person or team. Technical writers are perpetually backlogged. Product managers write release notes but don’t touch the API reference. The result is documentation that’s accurate on launch day and increasingly wrong every day after that.

The cost of stale docs is real but hard to measure. Support tickets increase. Developer onboarding takes longer. Integration partners get frustrated. Customers churn because they can’t figure out how to use features that work perfectly fine. Stripe has set the gold standard for documentation quality, but Stripe also has a dedicated docs team that most companies can’t afford.

The existing tools address parts of this problem. ReadMe, GitBook, and Mintlify provide great documentation hosting. Notion and Confluence store internal docs. But none of them solve the update problem. They’re containers. They hold whatever you put in them, and if you put in stale content, they serve stale content beautifully. Some AI writing tools can generate docs from scratch, but generating isn’t the hard part. Keeping docs current as the product changes is the hard part.

The Micro: A Stanford and Yale Team That Shipped at Cloudflare and Bond

Promptless is an AI agent that monitors your code changes, Slack conversations, and support tickets, then automatically updates your customer-facing documentation. The product works without manual prompting. When a PR merges that changes an API endpoint, Promptless detects the change and updates the relevant docs page. When a support ticket reveals a common user confusion, it suggests a docs improvement. When the UI changes, it can even regenerate screenshots automatically.

The feature set is surprisingly deep. It provides full citations, so every docs change links back to the PR, Slack thread, or ticket that triggered it. It maintains your team’s existing writing style and can integrate Vale linting rules for consistency. It learns from feedback, so corrections improve future updates. And it connects with your existing docs platform rather than requiring a migration.

Prithvi Ramakrishnan was VP of Product and Engineering at Bond, a fintech infrastructure startup that was acquired by FIS (NYSE: FIS) in 2023. He’s a Stanford CS grad from the AI track and a four-time International Mathematical Olympiad participant. Frances Liu led product, ML/AI engineering, and LLM operations at Cloudflare, OneSignal, and Graft. She went to Yale undergrad and started a PhD at Stanford with published LLM research before leaving to build. They came through YC’s Winter 2025 batch and have grown the team to five people.

The competitive set includes AI writing assistants like Writer and Jasper, but those are general-purpose and don’t integrate with engineering workflows. Mintlify recently added AI features. ReadMe has API documentation generation. Swimm does code-coupled documentation. But the specific niche of “monitor everything happening in your engineering org and automatically update external docs” is underpopulated. Testimonials on the site reference users at Bazel, Runpod, and Vellum, and one technical writer described it as a “solo tech writer’s godsend.”

The Verdict

I think Promptless is solving a problem that every growing software company experiences but few have prioritized fixing. Documentation drift is universal. The reason nobody has built a great solution before is that it requires understanding code changes, natural language processing, integration with multiple communication platforms, and the ability to write clearly in a company’s existing voice. That’s a hard technical problem, and the founding team has the credentials to pull it off.

The key metric is accuracy. If Promptless updates a docs page incorrectly, the trust model breaks immediately. Nobody wants an AI that introduces new errors into their documentation. The citation feature is smart because it lets reviewers quickly verify that an update is correct by clicking through to the source material. That’s the right design decision for building trust in an AI-generated output.

At 30 days, I’d want to see how many auto-generated updates get accepted without edits versus how many need corrections. That acceptance rate is the product’s core health metric. At 60 days, the question is coverage. What percentage of a company’s total documentation surface is Promptless actually keeping current? If it only handles API reference pages but misses tutorials and guides, the value proposition is limited. At 90 days, the signal is retention. Do teams keep using it after the initial novelty wears off, or do they revert to manual processes? The testimonials are strong, the integrations are broad, and the founding team has direct experience shipping AI products at scale. This is a tool I’d want my own docs team using.