Over 1 billion PDFs are created every day, but your agents still can’t read them reliably.
Today we’re releasing Parse 2.0, the most accurate document parsing API in the world.
Extend already processes millions of pages daily for leading AI teams like Brex, Mercury, Opendoor, Flatiron Health, and hundreds of others. Now, its even better.
Parse 2.0 is SOTA quality on RealDoc-Bench, our open source benchmark that measures agent success rate on real world docs that agents actually encounter in production.
We trained Parse 2.0 on 1M+ pages of the hardest documents seen in production. Here’s how it stacks up:
- #1 in healthcare, real estate, logistics, and financial services
- 95.7% agent Q&A accuracy on 581 docs (next best: 92%)
- 0.847 F1 on layout (next best: 0.759)
Give it a try today and build production-ready document agents with Extend.
In 1953, George and Barbara Bush lost their daughter Robin to leukemia. She was three years old. Barbara's hair turned white that year — she was 28. By anyone's expectation, mine included, their grief should have defined the rest of their lives.
Instead they made a decision: they would wake up every morning and be happy. It's an almost astounding level of mental discipline. The ability not to ignore your horrible loss but to say life is not over. I can choose how I feel.
It's not a coincidence that he became president and she raised one. More importantly, they stayed married 73 years. "You have two choices," she said decades later. "You can be happy or you can be unhappy. Just choose to be happy." This is what's key: that mindset wasn't influencing the outcome. The mindset was the outcome.
We tend to think of resilience as a trait people have, but it looks more like a series of small decisions, made unglamorously, that become habitual. Eventually, you'll even forget you're making them. The Bushes figured this out 70 years before the placebo sleep researchers did. Worth taking seriously.
It's funny watching all these consumer VCs doxxing how little they know about how even their own portcos achieved 'generational' scale.
The common thread: none of them have actually grown a consumer app, and they conflate spectatoring as an investor with actual experience.
A current best practice is that AI generated content should be consumed by you, not shared with others. This applies to code, bug reports, emails, and so on.
Here is my playbook for leading with transparency:
- Communicate the decision
- Share the rationale and tradeoffs
- Open the floor for questions
- Repeat as needed, reinforce it in the next few updates
It sounds simple. Most leaders under-communicate the "why". That's where alignment is won and alignment is what compounds into velocity and product quality.
We take a lot for granted.
Running the JFK promenade in GG park, morning walks at Crissy Field, sitting on the hill at Dolo. Beautiful spaces, and a lot of work went into breaking ground on them. Thousands of drawings, humans poring over those drawings, networks of subs and GCs making it all real.
Before dirt moves, someone looks at a plan and figures out what it takes to build it. How many plants, how much pipe, what it costs. This process hasn't changed in decades.
@JTLonsdale tried to get us to solve this problem years ago, and in my six years working with him he's normally a few years ahead of the curve (and we really should listen more sometimes). We explored it but the CV capabilities weren't quite there yet, and we never found the right team to pull it off.
Then we met Michael Ding building what we'd imagined, but much much stronger. Bobyard's AI reads drawings like a seasoned estimator. The value and time to achieving it is truly remarkable.
They started with landscaping, one of the hardest trades with the messiest drawings, and scaled quickly. Now they're expanding to framing, electrical, plumbing, and more.
Proud to lead their Series A.
While my main focus will be to be a dedicated student athlete, another driving factor of my decision is to be able to stay close to home and be a role model in my community. There's no better place to do that than Maryland 🐢🐢
Few people *care* more than @juliechentweets. She spent years setting the table at Retool and is now helping restaurants do the same thing better with Magic 🪄
After a year of quietly powering some of the best restaurants, we’re finally introducing Magic — the personalization engine for real world experiences.
Our first product, Loyalist, is live in hundreds of restaurants in 40+ cities like CARBONE, Momofuku, Le Bernardin, and COTE.
Incredibly inspired by @dtjourney and the rest of the Zingage team �� they’re building something special in healthcare and hiring talented folks in NYC!
@dtjourney and I are proud to announce @ZingageAI $12.5M in funding, led by Bessemer, TQ Ventures, South Park Commons, and WndrCo.
We are going to make home care the center of healthcare by removing the barriers that stand in the way of care delivery.
Here’s why: