Patlytics is an AI patent platform for law firms and in house IP teams. Its connected modules cover infringement, invalidity, freedom to operate, standard essential patents, portfolio analysis, claim construction, drafting, office actions and invention disclosures. A patent agent with more than 200 skills works across them, suggesting edits that users accept or reject line by line, and every claim chart cell cites its source.
Patlytics cites more than 130 million patents and 300 million literature items in its collection. Its August 2025 terms bar training on customer materials and carry an IP indemnity and a 12 month fee cap. Its trust center offers SOC 2 Type 2, ISO 27001 and ISO 42001 reports on request and lists OpenAI, Anthropic and Google Gemini among its subprocessors. Founded by Paul Lee and Arthur Jen and based in New York, Patlytics has raised about $65 million, including a $40 million Series B led by SignalFire on 8 April 2026.
Side by side on the same 15 capability axes and 12 legal signals: where the two part company, and what neither one publishes.
Vendors in the same categories as Patlytics whose graded profiles look most alike, with a line on what each product does.
Similar is not the same, so check that each one does the job you need. No vendor pays for placement.
Ordered by how closely the graded profiles match across the 15 capability axes, inside the same categories. See how each axis is graded in the methodology.
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
Every Patlytics module is model output. Its modules break claims into elements, find public material showing that products practice them, build cited infringement, invalidity and essentiality charts, and construe claims from the prosecution history. They also rank portfolios against competitors' products and draft applications and office action responses. A patent agent with more than 200 skills reasons over the results and packages them as emails, Word files or spreadsheets.
Patlytics says its models are fine tuned on more than 50 million public patents, and its subprocessors include OpenAI, Anthropic and Google Gemini. Without the models, Patlytics has no product.
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
Every claim chart cell in Patlytics cites its source, with pin cites to paragraphs, columns and figures in invalidity charts and to the public material behind each infringement match. Patlytics says it audits sources for relevance and shows color coded confidence indicators. It claims 99 percent or better verbatim excerpt accuracy and says its models are built to guard against hallucinations, without describing a test set or naming failure modes.
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
The Patlytics agent suggests edits line by line for the user to accept or reject, and Patlytics says every step ends with the user's approval. Claim charts mark each element as a match or for review, and freedom to operate output is framed for review before a product ships. The terms require customers to evaluate all output for accuracy and completeness before relying on it. Patlytics plans autonomous agents for drafting and research, and nothing states what any agent may complete without approval or where it stops.
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
Quinn Emanuel partner Yury Kapgan describes using Patlytics for infringement analysis, prior art, invalidity contentions, claim construction and charting. Patlytics announced adoptions by Foley & Lardner and Aspen Aerogels in November and December 2025. Asahi Kasei's Naoto Hori reports infringement investigations cut from months to hours, and $1,000 to $5,000 saved per detection report. An unnamed Am Law 100 practice group cut a 120 patent project from 100 hours to 20, turning a $47,500 cost into $38,000 of margin.
An unnamed biotech saves $5,000 to $7,500 per application, and Rivian, Polsinelli, Cleary Gottlieb, Wiggin and Dana, Repligen and Klein Tools also appear as customers. The case studies are undated and give the customers' own estimates without a method. Patlytics reports 55 percent of the Am Law 100 on the platform, against 40 percent in April 2026.
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
Training on customer materials is barred by Patlytics' terms, which also keep usage data free of customer materials and require deletion of confidential information after the agreement ends. Customer data is siloed and segregated by organization, encrypted, and reached through least privilege access. Patlytics' March 2026 checklist for practitioners covers Model Rules 1.1 and 1.6, the USPTO Rules of Professional Conduct and the section 102 prior disclosure bars.
That checklist asks whether work product is segregated by project and client matter, and Patlytics does not say that its own product segregates that way. The terms do not address privilege or how the model providers handle customer data.
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.
Patlytics' terms describe the service as a tool and supporting resource for patent work that gives no legal advice and no guidance on professional or ethical duties to clients. Customers and their users remain solely responsible for evaluating, verifying and applying professional judgment to all AI output. Authorized users must be the customer's employees unless an order form allows otherwise. Patlytics' March 2026 checklist for practitioners covers confidentiality under Model Rule 1.6 and competence under Model Rule 1.1. Limits by jurisdiction are not addressed.
Published governance over model behavior: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
Patlytics holds ISO/IEC 42001:2023 certification for its AI management system, and its trust center offers the certificate and full audit report through an access request. Controls listed there include a documented board charter and control self assessments. Patlytics claims 99 percent or better verbatim excerpt accuracy and says outputs carry source auditing and confidence indicators, but publishes no test method or result.
No accountable owner for the models is named, and nothing is disclosed about uneven output across technical fields, jurisdictions or claim types.
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
Data is encrypted at rest and in transit, with AES 256 for files in My Drive, and access follows least privilege. User history is kept for 90 days and can be deleted, and the terms require deletion of the customer's confidential information promptly after the agreement ends. Deletion of customer data on exit is among its listed controls, and nine subprocessors appear with roles and regions, all in the United States. Systems are monitored around the clock with on call staff in North America and Asia Pacific, and databases replicate to a secondary data center. Patlytics publishes no incident notification commitment or breach timeline.
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
Under terms updated 27 August 2025, Patlytics gives an indemnity against third party claims that the platform's underlying software infringes intellectual property. Output generated from customer inputs or specifications, and use in combination with other products, is excluded. Remedies are a license, a modification or replacement, or termination with a pro rata refund of prepaid fees. Liability is capped at fees paid in the prior 12 months, with indemnity obligations outside the cap and consequential damages excluded.
Patlytics warrants professional work by skilled staff and no malicious code, gives no warranty on the accuracy of AI output, and leaves reliance on output to the customer. A cybersecurity insurance document sits on its trust center behind an access request, and the terms promise no insurance.
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
Patlytics' plugin for ChatGPT Enterprise, built on Patlytics MCP, gives existing customers five read only tools. They cover semantic patent retrieval, patent lookup with full claims, non patent literature, the organization's private portfolios and its Patent Vault. Each user signs in with a Patlytics account, and results stay scoped to the organization. RPX Empower partnered with Patlytics in February 2026, and a Relativity integration has a waitlist after Relativity's investment. The agent exports Word and Excel files. No docketing, IP management or document management system is named.
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Patlytics runs as a cloud service on Amazon Web Services across multiple United States regions, with Auth0 and Datadog also in the United States. Its language model providers, OpenAI, Anthropic and Google Gemini, and its document parser, Reducto, are listed in the United States. Databases replicate in real time to a secondary data center. No customer choice of region, single tenant option or on premises deployment is published.
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
Patlytics' trust center offers its ISO 27001:2022 and ISO/IEC 42001:2023 certificates and full reports and its SOC 2 Type 2 report through an access request. A penetration test summary, a security overview, its cybersecurity insurance and its policies sit behind the same request. NCC Group and A-LIGN are named as its penetration testers. Controls listed openly include encryption key restrictions, firewall access limits, background checks, vulnerability monitoring and deletion of customer data on exit. Report dates and auditors sit inside the documents, which open after the request.
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
OpenAI, Anthropic and Google's Gemini are Patlytics' language model subprocessors, with Reducto for AI document parsing, each in the United States. Patlytics says its own models are fine tuned on more than 50 million public patents. No model or version is named for any module, and no notice is promised when a provider changes. Its ChatGPT plugin, launched on 17 September 2026, keeps zero data retention through the integration.
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
No price, tier or unit of charge is published. Patlytics' terms put fees in an order form, with annual renewals, 60 days' notice before a renewal increase, and no refund of fees paid. Patlytics states its value in savings instead, claiming claim charts that once took more than $30,000 of attorney time finished for a fraction of that.
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
Patlytics serves law firms and in house IP teams, and limits authorized users to the customer's employees unless an order form says otherwise. Modules cover infringement, invalidity, freedom to operate across jurisdictions, portfolio analysis, claim construction, drafting, office actions and invention disclosures. Standard essential patent work spans cellular, wireless, video and audio coding and memory standards, and life sciences drafting adds sequence listings and chemical structures.
Patlytics cites more than 130 million patents and 300 million literature items, while its invalidity page gives 70 million patent publications and 250 million non patent publications. No list of patent offices or date range is given for the collection.
The public pages on file for Patlytics, with the recorded signals each one supports and the date it was last read. Open any of them and check the reading against the record.
Primary Law Corpus Provenance, Good Law Verification
Read Oct 13, 2026
Prompt and Output Retention, Ethical Walls and Matter Segregation
Read Oct 13, 2026
Client Data in Training, Third Party Request and Subpoena Notice
Read Oct 13, 2026
Outside Counsel Guideline Readiness
Read Oct 13, 2026
Fabricated Citation Record
Read Oct 13, 2026
Court Disclosure Support
Read Oct 13, 2026
Bar Guidance Alignment
Read Oct 13, 2026
Billing and Fee Posture
Read Oct 13, 2026
Refusal and Uncertainty Behavior
Read Oct 13, 2026
Patlytics released a ChatGPT plugin built on the Patlytics Model Context Protocol server. IP teams can run concept level patent searches and query their firm's own patent data from inside ChatGPT.
A signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Can material a lawyer puts into this product be used to train a model?
The published terms prohibit training on customer content. Not a policy page, the agreement.
No customer materials go into training any AI model or system, Patlytics' terms state, and the usage data it may use to improve the service excludes customer materials. Its security page guarantees that customer data is not used to train or tune models. An FAQ answer on the same page says information is not used for training without explicit consent or necessity, a weaker wording than the contract.
How long does the product keep what a lawyer typed, and can that be set to zero?
A specific retention period is published and the customer cannot change it.
User history is kept for 90 days and the customer can delete it. The terms require deletion of the other party's confidential information promptly after the agreement ends, and deletion on exit is among the controls Patlytics lists. Whether uploaded documents and generated charts follow the 90 day period is not stated. The ChatGPT plugin carries zero data retention.
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
Patlytics keeps customer data siloed and segregated by organization, and access follows least privilege. Separation between matters or clients inside one firm is not described, although Patlytics' own March 2026 checklist asks vendors whether work product is segregated by project and client matter.
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Terms commit to notice where lawfully permitted. No transparency report located.
Under Patlytics' terms, a party compelled by a judicial or governmental order may disclose confidential information after giving the other party reasonable prior notice so it can contest the order. No transparency report is published.
Where does the law in this product come from, and does the vendor have the right to use it?
Sources are identified without stating the license or rights basis.
Patlytics cites more than 130 million patents and 300 million literature items, with prosecution histories preloaded, and its invalidity page gives 70 million patent publications and 250 million non patent publications. Standards documents for essentiality charts are indexed and kept current. No data source or license basis is named, and no office list or date range is given for the collection.
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
Patlytics preloads prosecution histories and builds claim construction from them. Nothing describes tracking whether a patent in an analysis is in force, expired, lapsed or invalidated, or a check of cited case law.
What does the product do when the answer is not in the corpus?
The product exposes a confidence or grounding score without an explicit abstention path.
Patlytics attaches color coded confidence indicators to its output and audits sources for relevance, and retrieval results show a relevance percentage for each reference. Infringement charts mark elements that need review. What the system does when no material supports a claim element, or whether it declines to chart, is not described.
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No court order, opinion or disciplinary record addressing fabricated or hallucinated legal citations produced by this product has been located as of the date shown. This is a statement about the public record on that one subject, not a finding about the product, and this signal is not a litigation history.
The AI Hallucination Cases database, Damien Charlotin's record of court decisions on hallucinated AI content, has no entry for Patlytics. Patlytics produces infringement and invalidity charts used in litigation, and drafts patent applications.
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
Patlytics' March 2026 checklist for practitioners cites the USPTO Rules of Professional Conduct, 37 CFR, Model Rules 1.1 and 1.6, the section 102 prior disclosure bars and state bar competence duties. It names no ethics opinion, and the terms say the output gives no guidance on professional or ethical obligations to clients.
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
Patlytics claims project time cut by up to 90 percent, and claim charts that once took more than $30,000 of attorney time finished for a fraction of that. Its studies cite $1,000 to $5,000 saved per detection report, $5,000 to $7,500 per application, and $38,000 of margin on one Am Law 100 project. How AI assisted charting or drafting should appear on a client's bill is not addressed.
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A current subprocessor or model provider list is published.
Nine subprocessors appear on Patlytics' trust center, all in the United States. Amazon Web Services hosts the service, OpenAI, Anthropic and Gemini for Workspace Studio supply language models, and Reducto parses documents, beside Auth0, Datadog, FullStory and Postmark. The SOC 2 Type 2 report, ISO 27001 and ISO 42001 certificates and reports, a penetration test summary and an insurance document are available through an access request. No data processing agreement is published.
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
Every cell in a Patlytics claim chart cites its source, invalidity charts pin cite paragraphs, columns and figures, and infringement charts mark each element as a match or for review. Users accept or reject agent edits line by line. Nothing records which model produced an output, or that an attorney adopted a mapping before it was served or filed.
Every grade and every signal on this index is drawn from public sources and dated. If a record is wrong, out of date, or missing an artifact the index did not locate, send the source and it will be reviewed and the record redated. Vendors are welcome to submit documentation. Nothing on this index is for sale, including a listing, a placement, or a grade.