Patent validity analysis with AI

Test a patent the way a challenger would. §102, §103, §112, §101, prosecution history and double patenting, claim by claim, with a 0–10 validity score.

A validity analysis examines a patent the way an opposing party would and estimates how well each claim would survive a challenge. One Transform builds its own prior-art record for the patent, then works through anticipation under §102, obviousness under §103, indefiniteness, written description and enablement under §112, eligibility under §101, prosecution history and double patenting, and gives the patent an overall validity score from 0 to 10. It is used before asserting or licensing a patent, when answering a demand letter, and when planning an inter partes review.

What you get

  • A summary with the overall validity score (higher means the patent is more likely to hold up), the most vulnerable and the strongest claims, and the risk on each ground.
  • An invalidity matrix: every target claim against every ground, each cell rated, and the dominant attack per claim with the references behind it and a recommended course of action, such as filing an IPR or a reexamination.
  • References for each claim, split into patents, non-patent literature and products, with the claim elements each one discloses.
  • §112 and §101 analysis, each finding tied to the claim language it concerns.
  • Prosecution and family: the prosecution events, the amendments and arguments that could create estoppel, and double-patenting exposure within the US family.
  • An opinion tab: an opinion-style write-up, an IPR strategy, the Graham factual inquiries and an analysis of the priority date.
  • A USPTO record panel, including any PTAB trials already filed against the patent.
  • Charts and exports: an Insights tab with the spread of claim risk and the estoppel picture, and Excel, PDF and Word exports with the same content.

How it works

  1. Ingestion. The patent, its priority date and its family are loaded. US prosecution history, continuity and PTAB records come from the USPTO, which matters for recent grants where other sources have little prosecution detail. You can also paste claims instead of a patent number.
  2. Claim parsing and construction. The target claims are split into elements and the key terms are construed. You choose the standard: Phillips for district court or broadest reasonable interpretation for the PTAB. Claim construction: Phillips vs. BRI explains why the choice changes results.
  3. Discovery. References are gathered from examiner citations, citation chains, the inventors' and the assignee's earlier work, classification searches, keyword searches, period vocabulary, conference literature and historical products. Only material before the priority date is kept, and the patent itself is never treated as its own prior art. You can add references you already hold.
  4. Pre-ranking. When discovery returns more references than can be scored claim by claim, they are ordered by closeness to the claims first, so the most promising art is examined in depth.
  5. Ground-by-ground analysis. Each reference is tested for anticipation, element by element. References that each disclose part of a claim are combined and tested for obviousness. §112 and §101 are assessed against the claim language and the specification. Prosecution events are read for estoppel, and continuations in the family are checked for double patenting.
  6. Scoring and opinion. Claim-level risk is built from the findings on each ground, and the overall score from the claims. The opinion and IPR strategy are written from the findings, including any PTAB history.

Dates, citations, family members and prosecution events come from patent office records. Disclosure findings, obviousness rationales and §112 and §101 assessments are model judgments, each tied to the passages and claim language it rests on.

When to use it

  • Before asserting or licensing a patent, to know which claims to lead with and which ones a defendant will attack.
  • When answering a demand letter, to see whether the asserted claims have a credible invalidity case.
  • When planning an IPR, to see which grounds and references are strongest in that forum.
  • In diligence on a patent you might buy, to price in validity risk.

For a reference set without the full ground-by-ground assessment, a prior art search is quicker. To check whether a patent blocks your own product, use freedom to operate.

Limits

  • The analysis sees published records only. It does not see the inventors' notebooks, sales activity that was never published, or secondary-considerations evidence that only the patent owner holds.
  • Obviousness, §112 and §101 are legal judgments. The analysis lays out the facts and the reasoning; it cannot decide them, and its reading of the case law should be checked.
  • Prosecution history is fullest for US patents. Foreign file histories are not analysed in the same depth.
  • The score is a research aid for ranking claims and grounds. It is not legal advice and not a validity opinion.

Frequently asked questions

What does the 0–10 validity score mean?

It summarises how well the patent is likely to hold up against the grounds analysed: higher means stronger. Each claim also carries its own risk rating per ground, and the score always links back to the findings, so you can see which ground drives it.

Which grounds of invalidity are covered?

Anticipation (§102), obviousness (§103), indefiniteness, written description and enablement (§112), patent eligibility (§101) and double patenting, together with prosecution-history estoppel. You can limit the run to the grounds and claims you care about.

Can I add my own prior art?

Yes. References you already have are added to the record and analysed alongside the ones the search finds.

Does it account for an earlier IPR?

Yes. PTAB trials on record for the patent appear in the USPTO panel and feed the IPR strategy and the opinion. Inter partes review: what a PTAB history tells you covers how to read them.

Is a validity analysis included in my plan?

Yes. It counts as one search, like every other search type. See pricing for the plans.

Try it on one of your patents

The fastest way to evaluate One Transform is to run a search on a patent you already know well.

7-day trial with 5 searches · No credit card · Work email required