Trarian Invalidity Score.
A defensible 1–10 invalidity score on every US patent — the number underwriters have been pricing assets without. Top drivers in plain English, the closest prior art surfaced, the historical invalidation rate for the band.
Trarian is an AI‑native research firm building the analytical infrastructure the patent market never had. We score, triage, and underwrite patents at scale.
For litigation finance, patent insurance, defensive aggregators, brokers, and IP‑backed lenders — anyone whose unit economics depend on telling the patents worth the work from the ones that aren't. The same defensible analysis on every patent that crosses the desk, in 24 hours.
Frontier models read the patent and the prior art. A calibrated statistical model grades. Patent‑qualified reviewers handle the judgment that matters. Same infrastructure, three ways it lands on your desk.
A defensible 1–10 invalidity score on every US patent — the number underwriters have been pricing assets without. Top drivers in plain English, the closest prior art surfaced, the historical invalidation rate for the band.
Rank a docket, an asset book, or a target's portfolio in an afternoon. Cut diligence down to the assets that earn senior hours — and pass on the riskiest decile before it eats the budget.
Prior‑art search, claim mapping, freedom‑to‑operate, contention briefs. Built on the same retrieval and ranking infrastructure that powers the score — senior reviewers on the judgment calls.
Two co‑founders who saw the same friction from opposite ends of the patent desk — and a founding engineer who's built the underlying systems before.
Former IP litigator at Kirkland & Ellis, representing technology clients in patent disputes. A decade as a go‑to‑market executive across technology startups before law school. Head of Product at Paxton AI alongside Mike, where the team scaled a generative AI legal research platform to multi‑million ARR.
IP strategy at McKinsey; data science at RPX, where he worked on the Rockstar portfolio acquisition. Head of AI at ZestyAI — built the computer vision pipeline scoring 200M+ properties for climate risk. CTO of Paxton AI. Author, "Your LLM is not a classifier."
Founding engineer at Paxton AI, where he built the platform that took the generative AI legal research product from zero to multi‑million ARR. A decade of full‑stack and infrastructure work across regulated industries — including investment banking integrations at Bank of America Merrill Lynch.
Mike came up through the licensing side at RPX, the world's largest defensive patent aggregator, working on the Rockstar acquisition — one of the defining defensive buys of the modern patent era. Jordan came up through the litigation side at Kirkland & Ellis, representing technology clients in disputes.
Different desks. Same friction. The reference you need to invalidate a patent exists somewhere. The probability an asset holds up is knowable. The question has always been whether anyone can deliver either in time for the deal.
At RPX, that friction decided which patents got bought, which got challenged, which got settled around. At Kirkland, it decided which invalidity theories made it into IPR petitions and which never got costed out. Better analysis, better outcome — on both sides of the table.
The root is Latin: trahere — to pull, to draw out, to extract. English drops the "h" — extrahere becomes extract. Add the suffix for one whose work is a particular thing — librarian, antiquarian, agrarian — and you get trarian.
One whose work is drawing things out.