Valuation & Funding
Knowtex has raised $5.5 million to date.
Caroline Zhang (CEO) and Jocelyn Kang (CTO) met at Stanford and founded Knowtex in 2022. Both have technical backgrounds. Before building the clinical product, they worked as medical scribes to see where doctors' time actually went, and they concluded that the note was only one link in a longer chain of chart review, ordering and follow-up.
Zhang studied economics at Stanford and earned an MS in computer science focused on AI. She did AI safety and natural language processing research, worked on AI policy in the Senate and at Schmidt Futures, and began biomedical research at the Salk Institute as a teenager.
Knowtex started in oncology, where founder research and time inside a community clinic exposed the hardest version of the problem. Cancer visits involve unfamiliar drug names, years of treatment history, overlapping illnesses and clinical trial options. The company now says it supports more than 200 specialties.
Product
Knowtex sells an AI assistant to the Department of Veterans Affairs, health systems and physician practices. It handles the paperwork around a patient visit: preparing the chart beforehand, writing documentation during the visit and lining up the tasks that follow. Physicians, nurse practitioners and other clinicians use it.
Before a visit, Knowtex pulls information from the electronic health record (EHR) through official integrations. Where those fall short, it uses "computer use" agents that click through the record's screens the way a person would. It extracts prior notes, abnormal lab trends, radiology impressions, pathology and genetic markers, and past therapies. The clinician then gets a summary, a suggested visit agenda and a list of trials the patient might qualify for.
During the visit, the clinician presses record, talks with the patient and presses stop. According to the founders, drafts appear within about 30 seconds. They include a visit note, suggested diagnosis and procedure billing codes, staged orders for labs or imaging, follow-up messages and, where a trial looks relevant, a message asking a research coordinator to start a workup.
The clinician reviews, edits and signs the drafts and approves orders; research staff confirm trial eligibility. The VA describes the same sequence in its rollout: the patient gives verbal consent, then the clinician reviews, edits and signs the note.
Knowtex calls itself a "frontier AI lab for healthcare." Its technical work centers on specialty-specific post-training, meaning additional training on medical tasks using foundation models from other developers. It also builds the software layer around those models: agents, EHR connectors, guardrails and evaluation tools.
Each customer gets forward-deployed engineers who map local workflows, such as how an oncology clinic orders chemotherapy labs. Where customer and partner data rights allow, clinicians' edits to signed notes feed back into customization and training. A clinician's signature is useful feedback, but it does not prove that a draft was clinically correct.
A February 2026 study in JCO Oncology Practice was coauthored by Knowtex's founders and the CEO of Astera. It compared 22 oncologists and 3 primary care physicians across a multisite community practice, split into an intervention group and a control group.
Measured against the control group, Knowtex users saw 3.35 more visits per day and billed $433.61 more per working day on average. Diagnosis codes per visit did not change, and there was no significant change in documentation quality. Those billings went to the practice, not to Knowtex.
KnowBench, a preprint written by the founders, covered more than 1 million signed encounters across 13 specialties. It found that clinicians kept 97.99% of the generated text without content edits. That figure measures how much draft text survived review, not time saved or clinical accuracy, and it covers only clinicians who kept using the tool. The founders also cite clinician feedback of roughly two hours saved per day.
Knowtex has begun offering what it calls a shared clinical brain: a clinician can inspect de-identified, aggregated cases resembling the patient in front of them and see which treatment plans other clinicians selected. Zhang also describes early research partnerships aimed at turning visit information into disease insights for research institutions and pharmaceutical companies; these capabilities extend beyond the documentation product, with clinical benefits still requiring validation.
Business Model
Customers pay Knowtex for software under enterprise contracts. The company's website lists no prices and directs buyers to contact sales, and health systems receive custom quotes. Under the VA's reported RFP terms, Knowtex bills per active clinician per month, with coding, orders and specialty support included in the base price. The unit price has not been published.
The VA awarded its ambient scribe contract on September 18, 2026. The contract sets a five-year maximum ceiling of $775.72 million, shared between the two prime awardees: Knowtex and Carahsoft, which resells Abridge. Individual task orders are competed, so the shared ceiling is not an allocation or revenue forecast for Knowtex. An earlier VA pilot carried roughly $15 million in contract value.
On September 25, Knowtex said it was deployed at 79 VA medical centers and more than 310 facilities. It reported over 7,000 clinicians using the tool and 88% adoption.
The buyer's case rests on clinician time. A doctor who finishes charts faster can see more patients, record complete billing codes or leave on time, while the organization pays a per-user software fee.
Knowtex's costs include fees to foundation-model providers for each encounter processed and cloud hosting. They also include security and compliance work and the integration and support engineers embedded with customers. Its compliance partner reports that Knowtex obtained a VA Interim Authority to Operate in 85 days. Knowtex describes its controls as aligned with FedRAMP Moderate, which is not the same as FedRAMP authorization.
Reusing EHR connectors, security work and specialty configurations across more clinicians could spread engineering costs over a larger customer base. That potential benefit depends on how much work Knowtex can reuse: extensive local customization, clinical validation and ongoing support could make each new health system expensive to launch and maintain.
Competition
Knowtex's most direct rival is Abridge, which competes with it for the same VA task orders and for the same health systems. The central question is whether hospitals will pay for an independent assistant that handles chart prep, orders and trial flags. The alternative is ambient documentation bundled into the record systems and physician apps they already use, increasingly at no extra cost.
Abridge
Abridge holds a position on the same VA contract through Carahsoft. Both vendors compete for task orders, with no fixed share of the shared ceiling allocated to either. Abridge says it works with both VistA and Oracle, the VA's legacy and new record systems. It also has distribution through Epic, which gives it a ready channel into large private health systems.
Knowtex's counter is breadth beyond the note, backed by forward-deployed engineers and integrations that work across different EHRs. That approach matters most if VA facilities and private hospitals value pre-visit and post-visit work enough to choose a smaller vendor.
EHR platforms
Microsoft's Dragon tools from Nuance and the ambient features Epic and Oracle build into their records systems have structural advantages. They appear inside screens clinicians already use and are bought through existing enterprise agreements. In February 2026, Oracle added drafting of lab, imaging, medication and follow-up orders to its Clinical AI Agent. That move narrows one of Knowtex's clearest differences.
The VA's move to Oracle Health raises the stakes. At sites that migrate, Knowtex must maintain its integration while competing with an EHR vendor that can offer its own assistant.
Free physician tools
Doximity launched a free Scribe for clinicians in July 2025, and OpenEvidence offers free Visits. These products reach individual doctors directly and turn basic note-writing into something clinicians expect at no charge. That puts pressure on per-user enterprise fees.
Knowtex competes on institution-wide security approvals, integration into the health system's record, and order and trial workflows that a department head can audit. The commercial test is whether those capabilities justify a separate purchase beyond a free note-writing tool.
TAM Expansion
Knowtex's documentation business draws on clinical operations budgets. Its access to patient histories, visit conversations and clinician-reviewed outputs creates opportunities to serve buyers beyond clinical documentation teams. The founders point to three such buyers: trial operations teams, drugmakers and research institutions, and health systems funding decision support. All three opportunities are early, and each depends on patient consent, data rights and clinical validation beyond what documentation requires.
Clinical trial operations
Research coordinators and trial sponsors spend heavily on finding eligible patients, often by reviewing charts by hand. Knowtex already flags possible trials during chart prep and routes coordinator messages. Combining record data with what a patient says in the exam room could support screening against protocol criteria, with coordinators confirming eligibility.
The founders cite one practice where trial matching surfaced 60% more patients than its prior baseline. That is a small, unverified company example. If the approach holds up, Knowtex could charge research budgets at cancer centers or sponsors, but no such pricing has been announced.
Pharma research and real-world evidence
Drugmakers and research institutions buy real-world evidence, meaning data on how treatments perform outside controlled trials, for R&D and post-approval studies. Visit conversations capture details that rarely reach structured fields, such as why a patient stopped a therapy or which side effects they tolerated. Knowtex describes these research relationships as early, not an established commercial business. Any revenue would depend on de-identification and on customers granting data rights.
Clinical intelligence and decision support
Knowtex has begun offering a shared clinical brain, drawing on de-identified, aggregated records to show clinicians which plans other doctors selected for "patients like mine." That product would sell into decision-support budgets rather than documentation budgets, a separate clinical technology purchase.
Expanding this early capability into a broader decision-support product would require evidence of clinical benefit beyond documentation studies. Knowtex would need to test for bias and unrepresentative data and secure appropriate rights to pool records across customers.
Risks
Federal concentration. Knowtex's largest deployment depends on VA task orders competed against Abridge under a shared ceiling that does not allocate future task-order spending to either vendor. Commure filed a protest with the GAO on September 23, and Vexcel/Microsoft filed one on September 28, which is due for decision by January 6, 2027; both were listed as open in docket checks in October and early November.
Platform bundling. Epic, Oracle and Microsoft can add order drafting and ambient notes to tools hospitals already pay for, which could shrink Knowtex's feature advantage. VA migrations to Oracle Health also require Knowtex to keep its integrations working on a competitor's platform.
Clinical and data safety. A wrong medication or staged order that a rushed clinician signs could undermine trust across an entire health system, and the company's evidence is mostly self-authored acceptance and billing data. The shared brain and research ambitions depend on consent, data rights and federal security reviews, and an error in any of those could also put existing contracts at risk.
DISCLAIMERS
This report is for information purposes only and is not to be used or considered as an offer or the solicitation of an offer to sell or to buy or subscribe for securities or other financial instruments. Nothing in this report constitutes investment, legal, accounting or tax advice or a representation that any investment or strategy is suitable or appropriate to your individual circumstances or otherwise constitutes a personal trade recommendation to you.
This research report has been prepared solely by Sacra and should not be considered a product of any person or entity that makes such report available, if any.
Information and opinions presented in the sections of the report were obtained or derived from sources Sacra believes are reliable, but Sacra makes no representation as to their accuracy or completeness. Past performance should not be taken as an indication or guarantee of future performance, and no representation or warranty, express or implied, is made regarding future performance. Information, opinions and estimates contained in this report reflect a determination at its original date of publication by Sacra and are subject to change without notice.
Sacra accepts no liability for loss arising from the use of the material presented in this report, except that this exclusion of liability does not apply to the extent that liability arises under specific statutes or regulations applicable to Sacra. Sacra may have issued, and may in the future issue, other reports that are inconsistent with, and reach different conclusions from, the information presented in this report. Those reports reflect different assumptions, views and analytical methods of the analysts who prepared them and Sacra is under no obligation to ensure that such other reports are brought to the attention of any recipient of this report.
All rights reserved. All material presented in this report, unless specifically indicated otherwise is under copyright to Sacra. Sacra reserves any and all intellectual property rights in the report. All trademarks, service marks and logos used in this report are trademarks or service marks or registered trademarks or service marks of Sacra. Any modification, copying, displaying, distributing, transmitting, publishing, licensing, creating derivative works from, or selling any report is strictly prohibited. None of the material, nor its content, nor any copy of it, may be altered in any way, transmitted to, copied or distributed to any other party, without the prior express written permission of Sacra. Any unauthorized duplication, redistribution or disclosure of this report will result in prosecution.