Knowledge infrastructure for AI. Mintlify powers docs for 20,000+ companies, reaching 100M+ developers a year. When humans or agents need to understand your product, they start with your docs. We make sure they find the right answers. We're hiring in SF at mintlify.com/careers.
External link for Mintlify
or
San Francisco, California, US
1 Post St
San Francisco, California 94104, US
Mintlify reposted this
Mintlify is adopting outcome based pricing. Every AI action now has a fixed price, and you only pay when we deliver a result. Your automations bills will drop by 70% on average, and all paid plans now come with unlimited usage for the Editor agent and Slack agent. Details below:
We are excited to roll out our new llms.txt architecture to all Mintlify powered docs. On large docs sites, agents navigate 52% faster while using 45% fewer tokens. The llms.txt file was introduced in 2024 to provide agents with a navigation map of a website’s content. However, if an llms.txt file gets too large, agents will receive large amounts of irrelevant information that bloats their context window, leading to more token spend and slower task completions. When we added /llms.txt after its initial proposal in 2024, we dealt with this by adding a 100,000 character cut off. But over the past two years, as the docs sites we power have grown, the limits of this temporary solution have started to be much more noticeable. To preserve coverage without expanding the root indefinitely, we rebuilt Mintlify's llms.txt generation architecture as a hierarchy of files, where the generator recursively identifies and moves complete navigation sections into child files through docs.json. On 200 tasks across ten docs sites, our improved version of llms.txt enabled agents to complete tasks 52.1% faster (363.6s to 174.2s), with 44.8% less tokens (1.74m to 960k). Find the full blog below:
Broke your docs? Mintlify now supports instant rollbacks. Previously, the only fix for a bad deployment was to revert and rebuild. For large docs, that could take up to 20 minutes, during which your users had to deal with inaccurate information. Now, you can restore an old deployment from the Mintlify dashboard in seconds. No more waiting on rebuilds. Instant rollbacks is live for all enterprise customers.
our talented designers have too much on their plates so i (the marketing intern) was forced to design this myself but we're hiring more designers and design engs! if you join us youll get to work alongside this phenomenal team: Dmytro Tovstokoryi, James Baduor, Fay Dakrouri, Emmanuel Dankyi, Sam (Kaishan) Ding
Mintlify reposted this
We spent years helping the world's best technical teams publish what they know. That's one side of the exchange: knowledge going out. The other side are agents trying to find and use that knowledge, which still runs on scraped pages and stale training data. Today we're bridging it with Mintlify Index: one universal search over the entire technical web, through an MCP or API. Index routes questions about products in Mintlify's corpus directly to publisher-maintained documentation. Questions outside it reach the wider web. Either way, the agent gets structured, citable context it can use to answer a question or plan the work. This follows naturally from what we already do. More than 200,000 documentation organizations run on Mintlify infrastructure, and Index carries that knowledge into the agent's context window. We benchmarked it across 150 implementation-planning tasks on 50 products hosted on Mintlify. A blinded pairwise judge preferred the Index-backed plan nearly two-thirds of the time, and the Index-backed agent finished 48% faster end to end. Mintlify Index is in public beta today. As agents write more software, documentation becomes infrastructure for how that software gets built.
LinkedIn is better on the app
Don’t have the app? Get it in the Microsoft Store.
Open the app