Trase Raises $107M for AI Agents in Healthcare Back Offices

Trase, a McLean, Virginia company building an operating system for AI agents, said on June 24 that it raised a 107 million dollar seed round led by ARCH Venture Partners, with participation from Red Cell Partners. That is an unusually large seed, and it brings the company's total funding to 117.5 million dollars. The money is not the interesting part. The signal is: AI agents are moving out of the demo and into the least glamorous, most rule-bound corner of healthcare, the back office.

Most AI headlines this year have been about chatbots and patient-facing tools. Trase is aimed somewhere quieter and, for a lot of health systems, more expensive: the administrative work that keeps a clinic running. Its bet, backed by real deployment data, is that the fastest return on AI in healthcare is not a smarter diagnosis, it is a faxed referral that routes itself.

What Trase actually built

Trase describes its product, Trase Origin, as an agentic operating system for regulated industries, healthcare and defense first. The company, led by chief executive Grant Verstandig and president Baskar Sridharan, says it runs hundreds of agents that can be deployed across a business in weeks rather than months. The platform runs in the cloud, on-premise, or at the edge, and it is built around what compliance teams ask about first: unified policy enforcement, continuous monitoring, and immutable audit trails. It is HIPAA and SOC 2 compliant, and it keeps a judgment layer that decides which calls a human still has to make.

The Duke numbers are the story

Since last fall, Trase has been running agents inside Duke University Health System's Division of Cardiology. The first production use case is boring on purpose: routing the more than 5,000 faxes the clinic receives every month, work that used to be hand-sorted by trained medical assistants and nurses.

According to Trase, the fax routing agent triaged those documents 7.1 times faster than people did, saved staff 1,395 hours a month, and freed up 285,450 dollars in annual staff capacity. The framing matters: those hours went back into patient care, not into headcount cuts. "We used to show up on Monday mornings to hundreds of faxes waiting to be routed," said Manesh Patel, MD, chief of the Division of Cardiology at Duke Health. "When we first turned on the Trase fax routing agent, it processed the lion's share of them by the end of the day."

Why the regulated back office is the whole pitch

The reason a fax-sorting agent is worth 107 million dollars is trust, not cleverness. In a hospital, an agent that misroutes a referral or drops a lab result is not a bug, it is a safety and compliance event. That is why Trase leads with sovereignty, auditability, and a human judgment layer instead of raw autonomy. "Agents are handling the onerous, immutable tasks that bog down highly trained individuals," Verstandig said. Robert Nelsen of ARCH put the investor logic plainly: the point is to augment capabilities in industries that are the most demanding for performance.

This is the shift worth watching. The 2026 enterprise AI story has moved from can the model do it to can you deploy it where the rules are strict and the mistakes are expensive. Healthcare is the hardest version of that test.

What this means for clinics and health systems

At PATech we build voice and agent workflows for exactly this kind of environment, so here is the practical read, not the press-release version.

Start where judgment is low and volume is high. Fax routing, eligibility checks, appointment scheduling, and referral intake are the right first jobs because a human can still verify the output and the savings are measurable in week one. Do not start with anything that touches a clinical decision.

Insist on the unglamorous features. A HIPAA business associate agreement, an immutable audit trail, and a clear human-in-the-loop rule for edge cases are not optional extras in a clinic, they are the product. If a vendor cannot show you the audit log, that is your answer.

Measure freed capacity, not just cost. Duke's own framing, hours reinvested in patients rather than staff removed, is also the honest way to sell agents internally. It keeps the people who know the workflow and puts them on higher-value work. For a related look at where this starts on the front desk, see our piece on AI voice agents and the missed calls that leak revenue first.

The sober version

One large seed and one cardiology division do not prove a category. The 1,395 hours and 285,450 dollars are Trase's own figures from an early deployment, not an independent audit, and back-office automation has a long history of pilots that never scale past one department. But the direction is clear enough: the near-term money in healthcare AI is in the paperwork, and the winners will be the ones who can prove, on an audit trail, that the agent did the right thing.