Extend

try it out in extractor: dashboard.extend.ai/studio/extract…

we just launched composer chat! iterating on document extraction schemas used to mean copy-pasting JSON into ChatGPT. it was error-prone, slow, and didn't always optimize for extraction. that's why we added a realtime mode to our composer agent. it has full context on your documents, existing schema, and best practices in Extend. just tell it what to change, and it applies patches directly to your extraction schema. we see customers using it to: - fill in field descriptions across your whole schema at once, or do broad refactors of nested fields - diagnose low accuracy and implement the fixes - optimize for latency and cost try it out in an extractor. link in comments.

try it out in studio: dashboard.extend.ai/studio

dark mode, now available in extend.

dashboard.extend.ai/studio try it here!

eval latency metrics are now live in extend studio! 🤩 see accuracy and latency breakdowns per run (p50, p90, p95) and track how they change across runs. this allows you to optimize for both simultaneously before hitting production, which is especially crucial for user-facing products.

myself and the founders of @mintlify, @greptile, @Vapi_AI, and @Cartesia ate 200,000+ scoville units of hot wings while answering off-the-record questions we talked about: - our worst sales calls - the closest our companies came to dying - a product launch that no one cared about - an early belief that turned out to be completely wrong turns out, every company has a graveyard of terrible first ideas and near misses

feb 2026 was one of our highest velocity months in history 🚀 some highlights include: - closed a 6-figure deal in just 10 days - launched our first billboard campaign in Montgomery station, SF buses, the 101 highway, and NYC subways - hosted 3 events between SF and NY, including founders on fire with vapi, greptile, mintlify, and cartesia shoot us a DM if you wanna join us, we're hiring aggressively across all roles! 🤩

our team looking good in front of our OOH campaign in Montgomery Station and the NYC subway 😎 if you spot us on your commute in SF or NYC, send us a pic!

@mercury case study: extend.ai/resources/merc…

we're super excited to share our work with @mercury! mercury processes $20B+ in monthly transactions. they tested Textract, foundation models, and other startups before choosing Extend. onboarding is the first thing a user sees on mercury's platform, and slow doc verification means drop-off. mercury hit sub-7s p90 latency with our light parser, and composer pushed edge case accuracy to 98% on their hardest docs. what we loved most about working with mercury's team is how much they care about the craft. customers like this make us better at what we do. link to the full case study in the comments!

we recently co-hosted a dinner with @browserbase. topics that made their way around the table: - using agents to automatically detect and fix slow database queries with Neon Postgres - running OpenClaw on a phone to automate personal and work tasks - one founder's 20-year bet on silver that's finally paying off📈 our one rule with dinners: make sure everyone has a great time. let us know if you want an invite to the next 🍷

rebrand spotted on the subway 💙 @ExtendHQ

API overview and migration guide: docs.extend.ai/developers/mig…

a new API version just dropped! here's what's new: 1/ no more polymorphic run endpoints - every endpoint now returns strongly typed responses per action 2/ synchronous endpoints mean no polling or webhooks for fast use cases 3/ robust polling utils with exponential backoff for async calls ... and much more! check out the API overview and migration guide in the comments 😁

spreadsheets with multiple tables per sheet are a nightmare to parse. that's why we recently shipped advanced Excel parsing! LLM-based layout detection that handles table separation and header recognition automatically.

try it out at dashboard.extend.ai/demo!

recently, we've been investing heavily in our parse API. customers have noticed. here's what our parse API volume looks like over the last month🚀

proud to share Extend's updated brand and website! we spent 100s of hours on it, and obsessed over every single detail multiple full redesigns thrown out, fonts swapped and swapped again, every animation tweaked until it felt right...we even locked ourselves in a room and white-boarded every single word on the page until it resonated why? prospects would see a demo and say something like "this is not what I expected based on your site", and they were 100% correct our product has changed so much in the past year, that we wouldn't even recognize the old version (new APIs, capabilities, entirely new categories of problems we now solve). Our old site didn't reflect any of that. we’re proud of how this turned out, and we hope it conveys the level of craft our team obsesses over in everything we ship check out the new site below and please share feedback!

@browserbase check it out here: browserbase.run/extend and let us know what you think!

we teamed up with @browserbase to automate one of the most tedious tasks in finance: collecting and processing expense receipts link below:)