Patent Continuation
Use AI to explore your patent specification for potential continuation claims
Review candidate subject matter with citations to your specification, then work with AI to draft and refine claims. Run the AI locally on your own machine, or use Claude, GPT and other models through your own OpenRouter account; OpenAI and Anthropic keys work too.
You start by editing a draft rather than from a blank page.
AI-assisted drafting. You direct the work, review the output, and remain responsible for the final submission.
For registered patent attorneys and agents, and professionals working under their supervision.
The download is Patent Practitioner Tools, made by Obviously Not, Inc. Nothing to activate and no account: run a model on your machine or use your own provider key, and every run reports what it cost. Why it is free.
Not ready to install anything? Start with the free course.
- Run discovery again keeps the embodiments you kept and lists new finds beside them.
- Each candidate embodiment includes a citation to the specification passage identified as supporting it, for you to verify.
- Draft first claims toward the one worth pursuing.
- And how it differs from the parent's claims, or that it only narrows one.
- How the embodiment relates to the parent's prosecution record: the allowed claims' limitations it builds on and the examiner's reasons for allowance, each checked against the record, or that nothing in the record was matched to it.
From potential continuation subject matter to a full draft claim set
Use AI to analyze the parent application's specification and claims for potential continuation subject matter, with its prosecution record beside the results. Choose the subject matter to claim, review and refine the AI's initial claims, then expand them into a full draft claim set, with your review and control at every stage.
Identify potential subject matter
Use AI to examine the parent application's specification and claims. Review candidate embodiments and claim directions, each with the passage cited as support for you to verify. With the parent's file history attached, each candidate also carries one sentence relating it to the parent's prosecution: the limitations of the claims as allowed it builds on, each placed where it entered the record (in the claims as filed or in a named round of prosecution), and the examiner's reasons for allowance. Every claim, round and quotation that sentence names is checked against the record.
Select a direction for drafting
Choose the embodiments or claim directions you want to develop, keep candidates for later, and dismiss the rest. Run discovery again explores more possibilities alongside the candidates you kept.
Review and refine the initial claims
Have the AI draft an independent claim and two dependent claims for your selected direction. Review the accompanying critique, edit the language directly, or instruct the AI to propose revisions. You determine whether the claims have adequate support and appropriate scope.
Expand, review, and export
When you are satisfied with the initial direction, have the AI expand the draft into a full claim set using your selected claim types and count. Review and revise the complete set, then save it as Text, Word or PDF for further review and preparation for filing; Changes since your download shows, word by word, what has changed since.
Below every step sits the Conversation panel, which takes requests in plain words, from an instruction for the next draft to a check asked for by name. Under any message of yours, Keep for every draft makes it an instruction that every later draft of the matter follows. Ask it where the papers discuss something, and it searches the specification, the parent's claims, the file history and the references in hand, and shows each passage in the paper's own words with its page, and a line for every paper saying whether it was searched.
- Next steps lists the findings waiting for your review and the words a claim uses that the specification's text does not hold.
- Text, Word and PDF save the drafted claims.
- Revise the flagged claims rewrites the claims the critique flagged; your originals are kept, side by side with each proposal.
- The critique's findings by weight: checked in the text, for your review, and notes.
- Claims lists the set beside the step, with each claim's count of findings.
In the conversation, ask for a set to be scored, and read the scores as a spread rather than a ranking: they separate gross differences, not near-equal drafts. A judge from the same model family as the model that wrote a draft is not counted toward that draft, and each score is labeled with its rubric version, so drafts are compared with drafts scored under the same version. Every step, captured.
What it gives a practitioner
Run the AI on your own machine
With a model running on your own computer, your application documents, the text read from them, and your drafts are processed on that computer and are not sent to us or to an AI provider. A model your Ollama server would send on to another machine, such as an Ollama cloud model, is refused. Use your own cloud provider account only when you choose to, matter by matter. On a Mac it is an app you double-click. If your firm's IT reviews software first, everything it will ask is on one page.
Where each limitation entered the record
Add the parent's Patent Center file history, one download at a time or several applications' downloads at once through Import from Patent Center. With your own USPTO Open Data Portal key, the app can get a published application's file history from the USPTO instead.
Every limitation of the allowed claims is placed where it entered the record, in the claims as filed or in a named round of prosecution, beside each rejection and the examiner's reasons for allowance. The parent is named with its issue date from its Issue Notification, or its notice of abandonment, beside the rule on copendency. Each date is shown with how long ago or how far ahead it is, and your docket stays the system of record.
Every quote checked against the page
It reads scanned pages on your machine and quotes the examiner's own words, each checked against the page it came from. A quotation shortened with an ellipsis is found part by part, in order, and shown with each cut marked; a word added, left out or replaced, or a different number, shows as not found. Each quotation opens its PDF at the page it is on. On real file histories a 9-billion-parameter model on a laptop found every rejection, and no model invented a reference. How that was measured.
Review AI-generated drafting critiques
Review the potential written-description and definiteness issues the app flags, together with suggested revisions. The critique assists your review; it does not determine patentability or establish compliance with §112.
Numbers you can check
The method behind every figure is published, so you can disagree with it.
What practitioners did
In 20 real parent and continuation families, the continuations dropped limitations added during prosecution 32% of the time, and limitations in the claims as filed 13% of the time. From public USPTO data: the three measurements, and a finding against the app.
The critique
The critique's antecedent-basis check was measured over 81 claims per model with Claude Sonnet 4.5 as the critic: it raised four antecedent-basis findings. Two named terms claim 1 already introduces, and the check dropped both; it set one aside with a line saying where its term was introduced, and confirmed one from the claims' own text. The seven antecedent findings examiners made in the same families all survive the check.
How long a run takes
Across 131 timed runs on 15 remote models, the median run took 78 seconds to a draft, and 84% finished in under three minutes. A local model on your own hardware is slower. The method.
The benchmark
50+ models were scored drafting continuation claims for 10 real patent families. Which one runs on your machine, and the full benchmark; for which model to choose in plain terms, the model guide.
Before you install
What does it cost?
It is free to download and use, with no account and nothing to activate. With a local model there is no cost per use; with a provider key you pay that provider directly, with no markup from us. More on cost.
What setup does it need?
To open it, none. To read scanned pages, an OCR engine (tesseract) on the same computer. To read what an Office Action states, and to draft, a model: one on your computer, through Ollama or another server that speaks the OpenAI API, such as LM Studio or llama.cpp; or your own key at OpenRouter, OpenAI or Anthropic. Try the sample opens with no model, because its application is already loaded; a model is needed from the first step, finding embodiments. More on setup.
What reaches a provider if I use my own key?
The text of the papers the model reads and, for a draft, what that draft is built from, sent to the provider you chose, under your key. More on what a provider receives.
Can I try it without client documents?
Yes. Try the sample opens a sample matter for either task: the public Office Action of application 18/541,216 for a reply, or a short published-style application for continuation claims. The public papers of 18/541,216 (its Patent Center file history, specification and allowed claims) can also be downloaded from /download and added yourself. More on the sample.
Does it work with Word?
Yes, both ways. Every paper can be added as a Word file, as well as a PDF or text, and struck-through text in one is read as deleted. Claims save as Text, Word or PDF wherever they are shown, and a reply downloads as Word, from Reply in Word or Report of the action in Word; notes on claims or a reply are never written into a download. More on the Word file.
Need help getting started? Write to [email protected]; it goes to a person.
The authorities, quoted and free to read
The decisions, statutes and rules of continuation practice, each quoted from its primary source and reporting what the authority held and how it bears on drafting.
Read the corpus, or take it in the order a family is actually built with the free course.
The same download, for Office Action responses
Patent Practitioner Tools also prepares replies: add the action, the current claims and the specification, see each rejection in the examiner's words with its deadline and the claims it applies to, then draft and check the reply. Its companion site, patent-actions.com, describes that side.
Start with the sample, not a client's file
Download Patent Practitioner Tools and press Try the sample, or try it on a public application from its file history. No client documents needed.