Anthropic's models sit at the center of how a generation of developers builds software, making documentation essential to the product itself.
<1 month
from contract to launch
10M+
average monthly documentation visits
1
writer to manage all Claude Code docs
Challenge
Anthropic needed documentation that could scale with the pace of Claude Code, a rapidly expanding product with daily ships, without pulling engineers onto infrastructure work.
Solution
Rather than standing up a documentation platform of its own, Anthropic chose Mintlify. Today, the team runs its own automations on top to pull reader signals from the Mintlify analytics API and fix documentation gaps.
Content that can keep up with the models
As the Anthropic team prepared to launch Claude 3.5 Sonnet in 2024, they needed a documentation platform that could grow with the rapid product adoption and influx of new users.
The solution had to perform at scale for a growing user base, support customization for the Claude brand, and enable developers to quickly find exactly what they needed—all without adding engineering overhead. Rather than engineer this infrastructure on its own, Anthropic chose Mintlify.
It immediately paid off, with the new docs experience launching in just three weeks. Tailoring the docs to match Claude's brand was fast, and Mintlify's docs-as-code workflow made updating content significantly easier for developers and collaborators alike.
It was really impressive to see how fast we could ship our docs with Mintlify.
Brian Krausz
Member of Technical Staff at Anthropic
The AI-native platform Anthropic uses
When Mintlify shipped llms.txt support across every site on the platform in 2024, it took a months-old proposal and put thousands of developer tools on it overnight. Anthropic's docs were among the first live, with no implementation work on their end. The two teams then developed llms-full.txt together, and that format was subsequently adopted into the official spec.
The result is agent-readable documentation since well before most of the industry even knew they needed to make their docs accessible to agents, or how. And no one at Anthropic has to maintain it. Built-in MCP servers and instant Markdown exports mean Claude's documentation plugs right into the AI tools its readers already use.
This also settles an issue every documentation team is now facing: agents need structure and completeness, while humans want a more opinionated, visual experience. But with machine-readability handled, Anthropic's team can write for humans, and Mintlify takes care of the rest.
One writer, millions of developers
Claude Code's documentation receives millions of visitors each month and is run by one person: Sarah Deaton, technical content engineer. She is pushing Mintlify and Anthropic tools to their limit finding ways to automate and augment her work. But one of the toughest parts of the job is knowing whether your documentation is helping users succeed, and how to improve.
In the past, it was difficult to collect and act on user feedback, even though it's the best signal you can get. It was so hard to sift through and took too much time
Sarah Deaton
Technical Content Engineer at Anthropic
Using the Mintlify analytics API, Sarah now runs scheduled jobs every few hours that pull in new feedback on pages, broken code-sample reports, and user feedback. An agent triages these and other signals like Slack mentions and GitHub issues for the open source Claude Code repository to determine whether they're user error, support issues, or valid documentation gaps that need to be addressed.
All of this information serves as context for Claude to improve the documentation. The labor of sourcing feedback, identifying valid content gaps, and drafting content is automated. The result is documentation that functions as its own feedback loop rather than a static surface.
The best fix isn't always in the docs
Not every signal coming through Claude Code's docs is a documentation problem. Some of it points to friction in the product itself, and now Sarah can take those fixes straight to the source. Feedback that would otherwise sit in a docs backlog reaches the product team directly, so user experience issues get solved rather than explained.
"I'm able to submit a PR to fix the actual product," she explains. "That short-circuits the loop of needing to document around a paper cut. You just fix the issue. Make it so that people don't even need to go to the documentation."
The bottom line
Anthropic moved its docs to Mintlify in 2024 to ship documentation fast without building infrastructure. Today, that decision is paying dividends. It shows up in the feedback loop that lets one writer run the documentation for Claude Code, turning reader signals into merged pull requests and product updates.
None of it would exist if the team had spent the last two years maintaining a documentation platform instead.