Knowtex

10/09/2026

Knowtex got a shout out in today. 🎉

It came in this week’s Lister Lowdown, the Forbes roundup of news about 30 Under 30 honorees like our cofounders Caroline Zhang and Jocelyn Kang.

The news: the VA named Knowtex one of two prime awardees on its Ambient Scribe Enterprise IDIQ to bring AI scribes across the VA health system.

Our scribe is already used by more than 7,000 VA clinicians at 79 VA Medical Centers. Behind every one of those numbers is a clinician spending less time on notes and more time with veterans.

So proud of this whole team. We’re just getting started.

Read it here: https://hubs.ly/Q04zPmm80
Our Ambient Scribe Enterprise IDIQ Announcement: https://hubs.ly/Q04zPmn00

This billion-dollar healthcare company is expanding. Plus, fintech Startup Valon Hits $2.3 Billion Valuation.

10/05/2026

We're heading to Miami for the Cornerstone Specialty Network Fall Summit, October 9 to 10! 🚀

Caroline Zhang, Nick Cassalia & Stacie Athman will be there all weekend. We'd love to hear how your practice is using AI today and what you want from it next.

DM Nick to schedule a meeting!

Thank you to our partners at Cornerstone for having us. See you in Miami.

10/05/2026

SF Tech Week is here and we're so excited to be part of it! 🎉

Our CEO Caroline Zhang is joining the founders of , and on Wednesday for Scaling Without Scaling Headcount, a candid roundtable on how fast-growing startups scale sales, infrastructure and support without adding headcount.

She's on the panel with Vaibhav Namburi, Spandana Nakka, Michael Buckwald and Yasser Elsaid, and we can't wait to hear them swap stories.

📍 Pump HQ, San Francisco 🗓️ Wednesday, October 7, 4 to 7 PM PT Only 35 seats, so grab yours: https://hubs.ly/Q04zcDFR0
Come say hi to the Knowtex team!

10/05/2026

Product Highlight: Smart Assist

Clinical notes shouldn't take more time than the care itself.
With Smart Assist, you tell it what you need in plain language, and it handles the rest:

✨ Reformat a note to match the way you like to document
✨ Make it concise without losing what matters
✨ Find and replace text in a few clicks
✨ Summarize files and chart data, including pasted or uploaded patient history
✨ Create a prompt that captures your formatting preferences, so every note is written the way you want
When you like the result, one click on "Use as Note" puts it in your note.

09/25/2026

Knowtex has been selected as one of two prime awardees on the VA Ambient Scribe Enterprise IDIQ.

Today, more than 7,000 VA clinicians use Knowtex across 79 VA medical centers and more than 310 VA facilities. Together they've saved more than 450,000 hours of documentation time, and that time goes back to Veteran care.

We earned VA's trust one clinic, one clinician, and one Veteran visit at a time. We're grateful for that trust and proud to keep building for the clinicians who care for our Veterans.

Read the announcement: https://hubs.ly/Q04yjZNF0

09/23/2026

Most AI benchmarks measure whether a model can do the work.

We wanted to know something harder: whether anyone kept it.

So we built KnowBench, and a metric called Effort Reduction. It counts the share of system generated work accepted by the responsible clinician under expert review. Deletions and rewrites count against us. Only what survives untouched scores.

Across more than one million signed encounters in 13 specialties, we score 97.99%, with per specialty results spanning 96.8 to 98.9%.

Full method in the paper.
https://hubs.ly/Q04yb96T0

09/15/2026

Caroline Zhang and Jocelyn Kang met as freshmen at Stanford studying AI and linguistics. They spent a combined year in scrubs as medical scribes, learning the work firsthand, and built the first version of Knowtex from what they saw.

They started in oncology, one of the most complex specialties in outpatient medicine, because a system built to hold up there would hold up anywhere. Oncology demands exactness in coding, staging, and longitudinal history that general-purpose AI was not built to deliver. They solved it at the model level, which is why the platform supports more than 200 specialties today without losing precision.

What makes us a lab is that we publish how we measure. Our new paper introduces KnowBench and our metric for Effort Reduction: the share of automated work a clinician accepts after review, measured only on signed encounters. Across more than one million signed encounters in 13 specialties, our systems score 97.99%.

Today we're introducing Knowtex as the first frontier AI lab for healthcare.

KnowBench Paper: https://hubs.ly/Q04xzbs80 Fierce Announcement: https://hubs.ly/Q04xzbs70
We're hiring at the Lab: https://hubs.ly/Q04xzbrm0

09/15/2026

Big news from our team: today we're launching as the first frontier AI lab for healthcare.

Here's what that means to us. Correctly coding an oncology chart. Understanding a Veteran's history across decades of care. Giving a physician more of the visit back for the patient. That's the edge of what AI can be trusted to do in real clinical work, and it's where we've spent years.

We started in oncology, one of the most complex specialties in outpatient medicine, on the bet that a system built to hold up there would generalize everywhere else. It does. The platform now supports more than 200 specialties.

What makes us a lab is that we build our own specialty-specific clinical models, our own anti-hallucination safeguards, and our own evaluation science. Our first benchmark, KnowBench, measures how much administrative effort we actually remove from a clinician's day.

It's deployed today across the U.S. Department of Veterans Affairs and 300+ other organizations nationwide.

Read the full announcement from co-founders Caroline Zhang (CEO) and Jocelyn Kang (CTO): https://hubs.ly/Q04xxLy00

08/11/2026

Look at the patient, or look at the screen.

Every clinician knows that tradeoff. The hours spent charting after clinic — "pajama time" — have become one of the leading drivers of burnout in American medicine.

Our new white paper is about removing that tradeoff inside the hardest environment in the country to attempt it: the U.S. Department of Veterans Affairs.

The argument underneath it: ambient AI is becoming the infrastructure clinicians build their day around, not a feature bolted onto an existing workflow. Which means it has to hold up in real conditions — inside CPRS, a decades-old EHR, at federal security requirements, for a patient population with disproportionately high rates of complex and chronic conditions.

VA evaluated more than 150 solutions in its national AI Tech Sprint. Knowtex was named a top 10 finalist, then a winner, then one of only two ambient AI vendors awarded the deployment.

The thing that made it work isn't a bigger model. A cardiology visit and a behavioral health visit don't sound alike and don't document alike. Specialty-specific workflow logic is the difference between a demo and a deployment.
VA's own published evaluation of the Kansas City site — 90 days, 18 primary care providers — found that all 18 wanted to keep using ambient scribe technology afterward, and most reported saving one to two hours of after-hours work.

One clinician: "Today was the first time I have been able to complete my notes and go home on time since I started working at VA 10 years ago."

Ten years.

Read the white paper: https://hubs.ly/Q04syB_m0

07/28/2026

Live at Veterans Healthcare Summit 2026.

Proud of our Knowtex team — meeting the clinicians, administrators, and federal health leaders doing the hard work of bringing AI into real clinical practice.

Our CEO Caroline Zhang took the General Session stage this morning to share what it actually takes to advance clinical efficiency inside federal healthcare: the wins, the walls, and the lessons worth stealing.

Come find us at booth #4. Let's talk about what real deployment looks like.