Mintlify

Today we’re releasing Mintlify Desktop in beta, one app for all of your docs and knowledge. Your team’s knowledge is split between internal runbooks, drafts, and the public docs your customers read. This means you are constantly context switching and digging around to find what you need. That's why we built Mintlify Desktop to give you a home for all of your knowledge. Available for download on Mac, Windows, and Linux. It comes with offline editing, and syncs when you’re back online. We’ve also improved collaboration on desktop and web by making it easier to see when you are collaborating on the same page with someone. Download link in comments 👇

Narration is now live for all Mintlify sites. Readers can listen to any supported page, with each sentence highlighted as it's read. No recording, no hosting, no re-recording. Available on Pro plans and above. Let your content do the talking.

We hope to see you next month in New York.

We pulled traffic logs across thousands of docs sites powered by Mintlify. In February, agents made up 20.8% of traffic, five months later in August that number jumped up to 66%. We then surveyed 329 documentation writers, engineering and knowledge teams. Three things stood out: 1. Docs built for agents fail 20x less. That’s 2.23 failed requests per task on HTML vs. just 0.11 with Markdown + llms.txt. 2. Confidence outweighs reality. 79% of teams rate their AI-readiness as ahead or average but only 7% have actually made every knowledge surface machine-readable. 3. Agents are already writing the docs. 83% of teams say AI drafts their documentation updates now, but only 9% are fully automated with no human in the loop, and 19% of teams have no one clearly responsible for keeping any of it accurate. Sarah Deaton at Anthropic shared: "The same page that misled one developer now misleads an unknowable number of agents." Your documentation stops being an after-thought and is quickly becoming your context layer: the surface area every agent that is trying to understand your product can then action on. More importantly, someone has to own it. Enter the era of the knowledge engineer. Someone whose job is making sure what agents read is accurate: reviewing drafts, closing the fragmentation gap, designing and maintaining context. Having a human in the loop is vital for system building that scales. Read the ungated research: https://lnkd.in/gaZYGqHt

Mintlify sites now natively support WebMCP The moment an agent lands on any docs page, it gets callable tools: search, navigate, read content, get skills directly No scraping or separate server required

AI agents have officially taken over humans as the number #1 readers of company documentation. Our State of Knowledge Report found that agents generated 66% of traffic across Mintlify-hosted documentation. Forbes covered what this shift means, and why knowledge engineering is becoming a critical function for AI-native companies.

AI agents are becoming the primary audience for company knowledge. Most companies aren’t ready. In August, Mintlify-powered sites recorded 257 million agent web requests compared with 131 million human page loads. Yet only 7% of surveyed companies have made every knowledge surface readable by agents. Our 2026 State of Knowledge Report examines what happens next: → Why machine-readable content alone isn’t enough → How better navigation reduced failed agent requests by 20x → Why 57% of companies lack a clear owner for AI-ready knowledge → How teams at Anthropic, HubSpot, and Decagon are adapting The takeaway: documentation is where the work starts. It can’t be where it ends.

Technical writers, content engineers, knowledge base managers, product educators, and support teams are increasingly taking on a new responsibility: knowledge engineering. The work is expanding from maintaining pages to designing systems that serve two very different audiences. Agents need exhaustive, structured information: parameters, schemas, constraints, and procedures they can reliably retrieve. Humans need context: why something works, which approach to choose, and what trade-offs matter. Knowledge engineers balance both. They connect product changes and customer feedback to content updates, establish authoritative sources, and make knowledge accessible wherever people and agents work.

Technical writers, content engineers, knowledge base managers, product educators, and support teams are increasingly taking on a new responsibility: knowledge engineering. The work is expanding from maintaining pages to designing systems that serve two very different audiences. Agents need exhaustive, structured information: parameters, schemas, constraints, and procedures they can reliably retrieve. Humans need context: why something works, which approach to choose, and what trade-offs matter. Knowledge engineers balance both. They connect product changes and customer feedback to content updates, establish authoritative sources, and make knowledge accessible wherever people and agents work.

We rebuilt our Automations experience to help you save time on repetitive documentation maintenance, so you can focus on building your product. Keep your docs in sync with code, draft changelogs, act on user feedback, maintain translations, fix broken links, and more. Go take something off your to-do list and let us know what you think: