Free filing-readiness screen
How were comparable applications treated?
Describe an invention and see what happened to thousands of comparable USPTO filings — how often they drew §101, §102, §103, and §112 rejections on first action, in plain English. No signup, no cost.
The description is required. The draft claim is only used as extra matching text and is never stored.
Watch an application
Filed already? We check the public file wrapper every day and email you a plain-English digest when a new document appears — what it is, what it means, and the reply deadline.
Your results
Check a description on the left — the verdict, what to do next, and the year-by-year first-action rates for the subclass appear here.
How it works
Three steps, no account, no cost.
1. Describe the invention. A few sentences about what it does and how it is different. A draft claim helps the match if you have one.
2. We match a CPC subclass. Keywords from your description are matched against the official CPC title list; you confirm the subclass, or pick your own.
3. You see comparable filings. First-action rejection rates for that subclass, from 2008 to today, built from public USPTO bulk data (PTOFFACT and the OACT weekly archives). Records can lag live filings by days — the weekly archives are published on Mondays.
A screen is statistics only — not a prediction for any one application, not a patentability opinion, not legal advice.
Common questions
Does a clear-looking result mean my invention is patentable?
No. These are rates across thousands of comparable filings. A low rate for a ground does not mean your application will avoid it, and a high rate does not mean yours will draw it. Only a patent attorney reading your claims can answer that.
Do you store what I describe?
No — a screen is not stored. We count screens only (how many, per day), never the description, the draft claim, or the subclass. A watch is different: it has to remember the application number and your email to check the file wrapper and alert you, so those are stored until you stop the watch — then they are deleted.
Is it really free?
Yes. Screening is free while we're testing, and watches are free. No signup, no account — the watch list is private to your email, unlocked with a 6-digit code.
How do application watches work?
When you watch an application, we snapshot its file wrapper first, so old documents never alert. Every day we check again and email you only if a new document appears, with the rejection grounds and the reply deadline in plain English.
What data is this built on?
Public USPTO bulk data: the PTOFFACT office-action dataset (2008–mid-2017), the OACT weekly office-action archives (2018–today), and the CPC title list for subclass matching. See "How accurate is it?" below for the coverage and the limits.
How accurate is it?
The numbers are reconciliation-checked, and the limits are stated here, not hidden.
Coverage: 10,350,026 office actions (4,384,532 PTOFFACT + 5,965,494 OACT), first-action rates for 640 CPC subclasses, 2008–2026 with no missing years. 2018 first actions are inflated by about 120,000 — applications whose true first action fell in late 2017, after PTOFFACT's cutoff — so read 2018's rates with that in mind.
Subclass match: 96.5% of applications join to a CPC subclass (2,111,665 of 2,188,039). The keyword suggestion ranks defensible neighbours, not the USPTO's exact main subclass — for example it suggests H04W for a wireless channel-prediction description whose true subclass is H04B. Always confirm the subclass; the stats follow your confirmed pick.
What the rates are: first-action rates only — the share of first office actions that raised each ground. The "allowed" figure is a proxy (the share whose last action allowed claims), not a grant rate. We do not show abandonment rates or time-to-allowance: the data has no outcome field for them.
Unresolved: §102 and §112 run lower in the OACT era than in 2015 (§102 45.7% in 2020 vs 52.0% in 2015). That looks like a real examination trend — §101 fell further, which matches the USPTO's 2019 eligibility guidance — but it is not proven, so we do not present the dip as a finding.