For patent counsel and licensing teams
Patent analysis that shows its evidence.
Claim charts, prior-art sets, validity and freedom-to-operate reviews, with every limitation tied to a passage you can check.
- 7-day trial with 5 searches
- No credit card
- Work email required
Evaluating for a team? Talk to us
- 01Linked databasePresent1
- 02Document nodesPresent1
- 03Directed citationsPresent1
Evidence
“Links represent the relationships between Entities, which can be directed or undirected.”
Product documentation, vendor website - 04Node importance rankPresent1
- 05Citation-based rank calculatorPresent1
- 06Random jump probability constantPartially present½
- 1Each limitation is checked on its own.
- 2Every finding quotes its source.
- 3Coverage ranks candidates; it is not an infringement finding.
Fig. 1 — Real analysis of US 6,285,999 B1, claim 1. Vendor redacted.
The platform
Nine searches, one workspace
Pick the question you need answered. Each search runs in the background and returns a ranked, evidence-backed result you can review, discuss and export.
Enforce and license
Find who may be practicing your claims, and where a portfolio is worth acquiring.
Defend and clear
Find the art that narrows a patent, grade its validity, and clear a product before launch.
Explore and plan
Map a field, find the closest patents to yours, and locate the open space.
How it works
From a patent number to a reviewable claim chart
No query syntax to learn. Start from what you have; the searching, reading and scoring run on their own.
- 01
Start from a patent or a product
Enter a patent number, upload a document, or describe the product in plain language, then choose the search.
- 02
The search reads the sources
Each source family is searched in parallel and every candidate is read in full before it is scored. Typical searches take 15 to 30 minutes and run in the background.
- 03
Review the evidence, export the work
Every result is ranked by claim coverage with the quoted evidence beside it. Export to Excel, PDF or Word.
Method
Scores you can audit
Candidates are ranked by how much of the claim the evidence covers, not by a model's overall impression. The coverage score orders results for review; it is not a finding of infringement, which needs every limitation.
- Element by elementEach claim limitation is rated present, partially present or not found against quoted evidence.
- Nothing is assumedA limitation without supporting evidence counts as not found and adds nothing to the score.
- Every finding is citedQuotes link back to the patent, paper or product page they came from.
- Validity by groundValidity is scored 0–10 under §102, §103, §112 and §101, claim by claim.
- What it cannot seeIt charts only what is published. Product details that are not public, paywalled specifications and source code are out of its reach.
- 13 present
- 13 × 1 = 13
- 3 partially present
- 3 × ½ = 1.5
- 0 not found
- 0 × 0 = 0
- Coverage of 16 limitations
- 14.5 ÷ 16 = 90.6%
Your data
Built for confidential work
Invention disclosures, litigation targets and portfolio strategy are confidential. Here is how the platform treats them.
Completing a security questionnaire? Contact us
Private to your account
Your searches, uploaded documents and results stay in your account and are never visible to other customers.
Not used for training
We do not train models on your searches, documents or results.
Named model providers
Professional runs searches on Google Gemini and OpenAI; chat runs on OpenAI. Search terms also go to patent and literature databases and a web-search service.
Your own model keys, if you prefer
On the Bring Your Own Models plan, searches run on your Anthropic, Google or OpenAI account under your agreement with them.
Work accounts, one session
Sign-up requires a work email, sign-in is protected against bots, and each user has one active session at a time.
Pricing
Start with 5 searches, free
Start with a 7-day trial that includes 5 searches. No credit card required.
Each claim limitation is rated present, partially present or not found against a quoted passage, and the score is computed from that coverage, not from general similarity. The method has limits: it can only chart what is published, so a product with no public documentation cannot be analysed, and results are a starting point for a practitioner's review, not a conclusion.
No. One Transform is a research tool. Its charts, scores and reports support a practitioner's judgment; they are not legal opinions, and they should be reviewed before anyone relies on them.
On the Professional plan, searches run on Google Gemini and OpenAI models through their APIs, and the chat assistant runs on OpenAI. On Bring your own models, searches run on your own Anthropic, Google or OpenAI account. We do not train models on your data.
Your searches, uploads and results stay in your account and are never shown to other customers. We do not train models on them. On Bring your own models, searches run on your own model provider account. See the Privacy Policy for retention and other details.
One run of one search type on one patent or product. Searching three patents for infringement counts as three searches. Chatting about results does not use searches.
Create an account to start a 7-day free trial with 5 searches included — no credit card required, or contact us and we'll walk your team through it.
Your results stay in your account. To run new searches, choose a plan; nothing is charged unless you subscribe.
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