Heartflow Sues Cleerly Over Cardiac AI: Six Patents, a Former Insider, and a Warning for Every AI Startup

On April 13, 2026, Heartflow, Inc. (Nasdaq: HTFL) filed a patent infringement lawsuit against Cleerly, Inc. in the U.S. District Court for the Eastern District of Texas. Six patents. Priority dates running from 2012 to 2018. Three accused Cleerly products: Ischemia, Plaque Analysis, and Compare. The complaint seeks permanent injunctive relief and damages. It also tells a second story, a trade-secrets narrative about Cleerly’s founder Dr. James Min, who the complaint alleges “secretly created Cleerly in 2017 while he still worked for Heartflow,” gaining what Heartflow calls “intimate access” to its cardiovascular diagnostic technology, trade secrets, and confidential business information before incorporating a competitor without telling his then-employer.

Two companies using AI to analyze coronary CTA images and help cardiologists detect coronary artery disease. One says it spent since 2010 building 160 million annotated CTA images, 600 granted patents worldwide, 200-plus clinical studies covering 365,000 patients, and adoption at more than 1,800 institutions. The other is a well-funded Denver-based rival. The case will be litigated in the Eastern District of Texas, Judge Gilstrap’s backyard and the single most active patent court in the country. Heartflow picked its forum.

Why this case matters beyond cardiology

Most AI patent disputes so far have been about training data, model outputs, and whether generative systems infringe copyright or read materials they had no license to read. Heartflow v. Cleerly is different. This is classic utility patent litigation over how an AI system works, applied to a regulated medical device market. It is what AI patent litigation looks like when the underlying technology is stable enough to protect with claims, when the market is big enough to be worth fighting over, and when one company watched an insider walk out the door with knowledge.

Health tech is the first industry where the math works. Cardiovascular disease is the leading cause of death globally. Noninvasive AI-driven imaging analysis is a multi-billion-dollar market with two or three credible competitors. Heartflow is a public company with a real patent portfolio. Cleerly has raised hundreds of millions. Both have enterprise customers on long contracts. Injunctive relief in this context is not theoretical. It could remove a product line from hospitals that built clinical workflows around it.

Expect the same pattern in legal tech, compliance AI, underwriting AI, and any vertical where the incumbent has invested in models, curated data, and peer-reviewed evidence, and where a rival hires someone out of the incumbent.

The insider-founder problem

Every GC should read paragraph one of the Heartflow complaint and think about their own company. The allegation is that Min incorporated a competing entity while still under contract with Heartflow, with access to its technology and confidential information, and that the competitive product later infringed six patents. Whether the allegation is accurate is for the court to decide. The pattern is not exotic. Technical founders leave incumbents all the time. The question is what evidence the former employer preserved, and what the departing employee signed.

Three documents decide these cases before discovery closes. The employment agreement (assignment of inventions, confidentiality, non-solicit, cooperation obligations). The separation agreement (return of property, post-employment obligations, surviving covenants). The exit interview log (devices returned, accounts shut down, representations made). If any of those are weak or missing, the trade-secret claims get harder and the patent claims carry all the weight.

For the departing founder, the playbook is equally clear. Clean-room development. No use of former-employer materials. Independent invention documentation. Freedom-to-operate opinions from outside counsel before launch. A new employer or new venture that does not begin work until the old employment relationship is fully terminated.

Eastern District of Texas, with six patents and a priority date from 2012

Heartflow filed in Marshall and Tyler divisions of E.D. Texas because the district moves patent cases faster than most federal courts, has a bench with deep patent experience, and has historically produced plaintiff-friendly outcomes. The court has local patent rules with strict deadlines for infringement contentions and invalidity contentions. A defendant usually has to commit to its invalidity theories within four to six months. That compresses the case and favors the plaintiff who prepared before filing.

Six patents with priority dates from 2012 to 2018 is a serious portfolio. Pre-AIA and post-AIA filings will require different prior art analysis. The accused Cleerly products were launched after Heartflow’s filings in most counts, which simplifies the infringement narrative. Expect Cleerly to challenge validity in inter partes review at the PTAB within 12 months. Expect Heartflow to oppose any stay of the district court case pending IPR, especially given the forum.

Damages calculations in regulated medical device markets get large. Royalty bases can include per-scan fees, per-patient fees, and subscription revenue from hospital customers. Lost-profits theories apply where the companies are direct competitors. Heartflow is asking for damages and an injunction. The injunction is the lever.

Priority dates matter because AI priority is about to become a standard issue

The 2012 priority date in this portfolio is significant. Most companies selling AI products today were founded after 2015 and filed most of their patent applications after 2018. If an incumbent has a legitimately early priority date on a foundational method claim, that incumbent has a weapon the market forgot about. Patent portfolios built during the first AI wave have been sitting quietly while the second wave commercialized. Heartflow’s complaint is an early signal that those older patents are about to matter.

Every AI company with meaningful IP should do three things this quarter. Pull the full patent landscape in your subcategory, not just your own filings. Find every patent with priority dates earlier than your founding year. Run a freedom-to-operate analysis against them. If you find a cluster of foundational claims held by a public competitor or a patent assertion entity, brief the board before the complaint hits the docket.

What to do now

If you are in-house at a company deploying AI in a regulated vertical (healthcare, finance, insurance, legal, compliance), pull your freedom-to-operate memos. Confirm they cover every product feature you have shipped since the memo was written. Feature drift is the single most common cause of inadvertent infringement.

If you have a technical co-founder or VP of engineering who came from a well-funded incumbent, review their employment, assignment, and separation agreements with the old employer. Confirm your onboarding documented independent invention. Confirm your development environment has never touched the old employer’s code or data. Preserve the evidence now, not after a letter arrives.

Update your IP indemnities in customer contracts. The standard “we indemnify for third-party IP claims” is fine until the first injunction motion lands and your customers start asking what the backup plan is. Build the backup plan. Know which features you can remove, replace, or sublicense, and how long each would take.

For patent prosecution, file on what you actually built, not what you hope to build. Continuations exist for a reason. Keep families open. Priority dates compound in value every year AI commoditizes.

The Heartflow complaint is the first of a wave. The next will be in legal tech, document review, underwriting, or fraud detection. Every GC at an AI company should read it this week and run the same three checks on their own operation.