About Tremont AI
Tremont AI builds foundation models and software that help pathologists read preclinical drug safety studies. TRACE is the vision encoder that sees the tissue, and ToxScribe is a vision-language model that describes what it sees. Agents then link those findings to other study data, such as clinical chemistry, organ weights and gross observations. Tremont Studio brings all of this into one workspace, where a pathologist reviews the evidence and signs off on the study. The products are built for pharma and CRO pathology teams.
Tremont AI is based in Boston, Massachusetts. It is a spinout of the Mahmood Lab and uses technology exclusively licensed from Mass General Brigham. Luca Leon Weishaupt is co-founder and chief executive officer, and Long Phi Le is a scientific co-founder. Tremont AI came out of stealth on September 29, 2026. On October 27, 2026, it announced a research collaboration with AstraZeneca. The project covers building and testing AI models for preclinical safety assessment.
Funding Rounds (SEC Form D)
| First sale | Filed with SEC | Amount sold | Investors | Avg. check | Terms | Filing |
|---|---|---|---|---|---|---|
| Feb 9, 2026 | Nov 4, 2026 | $2,000,000fully sold | 13 | $153K | EquityRule 506(b) | Form D ↗ |
First sale is the date the first investor in the offering committed to invest. Companies must file Form D within 15 days of that date, so filing dates run later.
Why Investors Backed Tremont AI
Tremont AI builds foundation models and a collaboration platform for pathologists who review preclinical drug safety studies. The Boston company is a spinout of the Mahmood Lab and uses technology exclusively licensed from Mass General Brigham. It came out of stealth on September 29, 2026. On October 27, 2026, it announced a research collaboration with AstraZeneca.
The problem it targets
Tremont AI's case is that toxicologic pathology holds up drug programs. The company wants preclinical safety work to be faster and more certain, so the right medicines reach patients. A study records tissue slides alongside clinical chemistry, organ weights and gross observations. The models read the tissue and weigh it against everything else the study recorded. Pathologists still make the call.
What the product does
TRACE is the vision encoder that reads tissue images. ToxScribe is a vision-language model that describes the tissue morphology it reads. Agents then connect those findings with other study data, such as clinical chemistry, organ weights and gross observations. Tremont Studio is a shared workspace for reviewing whole slides, annotating them and working together in real time. The home page shows a sample study with 225 slides and six organs, split into dose groups. Customer study data is not used to train the models unless the customer agrees in writing.
Progress so far
Tremont AI came out of stealth on September 29, 2026, with a platform built on technology exclusively licensed from Mass General Brigham. On October 27, 2026, it announced a research collaboration with AstraZeneca. The project covers building and testing AI models for preclinical safety assessment. The company now works with a small number of pharma and CRO pathology teams, who get early access and evaluate the tools on their own studies.
We wrote this summary from public statements by Tremont AI and its investors. It is not investment advice.
Tremont AI in the News
Leadership Team
Executive officers and directors listed on Tremont AI's Form D filed Nov 4, 2026.
Sources
- SECTremont AI, Inc. filing history
- CompanyTremont AI team page
- CompanyTremont AI newsroom
- CompanyTremont AI privacy policy