Quickstart

This page is a first run: it takes you from a blank account to a real result. You will start a research database, put one patent and one product into it, and let Patalyze compare them. The comparison is a mapping: it breaks the patent's claims into their elements, judges each one against the product, and scores how close the read is. That score, sitting in a table where the two meet, is the result everything else in Patalyze builds on.

By the end of this page

You will have a research database holding one patent and one product, a mapping Patalyze generated between them, and a colored score you can open to see the evidence claim by claim. About five minutes, start to finish.

Pick a direction#

Most work in Patalyze runs in one of two directions, and the direction decides which side you start from. Pick the one that matches your question. Each card opens the workflow in the app, where the steps below play out.

Freedom to Operate

Start from your product and find the live patents it could infringe. (opens the app)

Infringement Detection

Start from your patents and find products on the market that practice them. (opens the app)

Not ready to map anything yet? You can also just explore the patent corpus and the product catalog every research database draws from. Search patents to query millions of global filings, or search products to find and crawl real-world items into structured features.

Search patents

Query the global corpus of patents.

Search products

Discover and deep-crawl real-world products.

Run your first analysis#

These five steps walk the full path, from an empty database to a scored result you can open. They read the same in either direction; where the two diverge, the step says so.

Start a research database

From your library (the home screen listing your databases), create a new one and choose a workflow: Defensive / Freedom to Operate or Offensive / Infringement Detection. The workflow builds the database around your question, seeds a starter search, and lays out the pages you will read the result on. Give it a name that pins down the question, so the verdict stays easy to trust later. One question per database keeps every conclusion clean.

Add the side you are starting from

Defending a product? Add it from the Add menu: search the catalog, crawl a product-page URL, or upload a datasheet. Asserting a patent? Enter its publication number, or query the corpus to round out the family. Either way, this is the anchor of the analysis: the fixed side that everything else gets measured against.

Gather the other side

Now bring in what you will compare against. In freedom-to-operate that means the patents your product has to clear; in infringement detection it means the products that might practice your patent. The search seeded at step 1 gives you a head start. Aim for recall: it is better to pull one extra than to miss the one that matters, since the scoring is what separates real risk from noise.

Let Patalyze map them

Mappings appear on their own. Once a product has learned its features and a patent is in the database, Patalyze creates a mapping for each pair in the background. Every mapping breaks the patent's claims into their elements, judges whether the product practices each one, and rolls the verdicts into a per-claim score and a risk band. Give it a moment; the scores fill in as the work finishes.

Read your first result

Open the table the workflow laid out. Where a patent and a product meet, a single colored cell carries the claim score; Critical and High are the close reads. Click a cell and open its explanation to see the full mapping, every element with the evidence behind it. When the analysis is done, Patalyze also writes its executive summary, a note with the conclusion ready to share.

That is the whole loop in miniature: a research database, a patent, a product, and a scored mapping between them. Real analyses are the same shape at scale, with more on each side and dashboards to chart the field. To see the full path for your direction, read Freedom to Operate or Infringement Detection.

Run it from an AI client

Every step here is also available to an AI agent through MCP. A connected client can start the research database, add your patent and product, and read the mapping back, all without opening the app. See the Claude and ChatGPT guides to connect one.