P
PQAI

PQAI, short for Patent Quality through Artificial Intelligence, is an open source prior art platform with a hosted web tool and a paid API. Sam Zellner leads the project, which grew out of work by the Georgia Intellectual Property Alliance and AT&T. GreyB contributed the retrieval algorithm, and InspireIP manages the application. The models are trained on historical patent examination data from the United States Patent and Trademark Office.

Inventors, patent attorneys, corporate intellectual property teams and technology transfer offices describe an invention in plain language and get ranked patents and other literature from a collection of more than 100 million patents.

The web tool adds a Concept Extractor, a Keyword Suggester and a predictor that maps a description to Cooperative Patent Classification codes. PQAI+ returns up to 500 results per query with full text excerpts and export, against 20 on the free plan. Software teams call the REST API to put ranked prior art into an existing disclosure or review system, and the code is available for a local installation. PQAI keeps no record of what users type and sends none of it to a training pipeline or third party AI engine.

Commercial research, including client work and law firm research, is allowed on every plan, Free included. PQAI+ costs $20 a month or $216 a year, and building PQAI into a product requires an Enterprise API plan listed at $700 a month.

Capability grades

All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.

AA on AI CentralityThe artificial intelligence is the product. Remove the models and there is nothing left to sell.

AI Centrality

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.

The core service is a semantic retrieval model that ranks material from a description of an invention or a patent question, across more than 100 million patents and nonpatent literature. It suggests Cooperative Patent Classification codes, extracts technical concepts and proposes terms for the next query. A user can start in ordinary technical language and then apply filters to narrow the results, and PQAI+ adds full text excerpts and faster processing.

The model is trained on examination data from the United States Patent and Trademark Office. The same capability runs in the hosted web tool, through the API and in open source components that teams can install themselves.

Source: Vendor Published
BB on Citation Accuracy and Hallucination DisclosureGrounding is real and documented, with linked primary sources and a described retrieval method, short of published accuracy figures an outsider can test.

Citation Accuracy and Hallucination Disclosure

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.

Results point to patent publications and other material a user can open and inspect, and the patent data endpoint returns claims and bibliographic details. PQAI tested its model in a public prior art competition to see whether it would surface references experts had identified. A document's relevance rank does not make it prior art for a given claim, and the interface lets users review dates and publication details. No general error rate exists, and nothing validates citations produced by an outside assistant that calls the API.

Source: Vendor Published
CC on Autonomy and Oversight ModelAutonomy is claimed and oversight is asserted without a mechanism. Human in the loop appears as a phrase rather than a described control.

Autonomy and Oversight Model

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.

A user enters the invention description and picks the options. PQAI returns ranked results and can suggest concepts, terms or classification codes to refine the next query. A technology transfer office can call the API from its disclosure system and attach the returned records for staff and outside counsel to review. The software supports early screening and source retrieval. It makes no patentability decision, changes no disclosure record outside the host application and files nothing, so the reviewer decides how to use the results.

Source: Vendor Published
BB on Operational and Outcome EvidenceReal deployment evidence with substance, short of full attribution or measurement: a named customer without figures, or figures without the named customer.

Operational and Outcome Evidence

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.

Akshay Harlalka built a life sciences patent tool on the PQAI API during an AI fellowship. An unnamed technology transfer office reports reviewing 30% more invention disclosures in the same period and handling 15 extra submissions in the first month after embedding the API. PQAI also ran its model against prior art references chosen in a public competition. These accounts show adoption and the scope of the workflow, with no general accuracy or recall figure for patent research.

Source: Vendor Published
BB on Privilege and Confidentiality PostureSubstantive published commitments on confidentiality and training use, short of the full picture: commonly silence on segregation between users or matters, or on what the underlying model provider may retain.

Privilege and Confidentiality Posture

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.

What users type is not stored on the back end, according to the hosted terms, and the security page says it enters no training pipeline or third party AI engine. Requests travel over an encrypted connection. A team can also run the open source software on its own infrastructure and control access itself, an option suited to an invention that is still confidential before filing. There is no privilege undertaking for submitted material, no ethical walls for a hosted law firm account, and no statement of which commitments apply when an outside AI assistant calls the API on a user's behalf.

Source: Vendor Published
CC on UPL and Professional Responsibility PostureA boilerplate disclaimer sits in the terms while the marketing describes the product in advice terms, or the intended audience is left ambiguous.

UPL and Professional Responsibility Posture

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.

Inventors and IP practitioners use PQAI for early research, classification and review of documents to weigh when evaluating an invention, and client work is a permitted use on every plan. Results are records and references for a person to review. In its benchmark write up, PQAI explains that dates and family relationships decide which records belong in a real invalidity analysis. No bar guidance applies to the workflow, and nothing routes a legal question to a patent attorney or agent.

The tools serve retrieval and investigation before a professional reaches a conclusion on patentability or legal status.

Source: Vendor Published
BB on AI Governance and Bias DisclosureA published governance framework with real substance, short of testing results or a named owner.

AI Governance and Bias Disclosure

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.

The project releases its source code, explains how examination data trains the model, and has written up a retrieval benchmark using references chosen by experts in a separately judged public competition. Users adjust filters and inspect every returned document, and technical users can study and test the open source components directly. There are no bias measurements across jurisdictions or technical fields, no evaluation schedule for updates to the hosted ranking and no escalation policy for inaccurate results. The benchmark speaks to retrieval behavior within the scope of that competition.

Source: Vendor Published
BB on AI Safety and Data StewardshipSubstantive published policy covering most of the ground, short of the full set: commonly no named subprocessor list or no stated incident practice.

AI Safety and Data Stewardship

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.

The hosted platform keeps no copy of user requests on its servers and sends none to a training pipeline or third party AI engine. API traffic runs over HTTPS with token authentication. Account information used for subscriptions, billing and support is handled apart from the confidential content of an invention description, and the free plan needs no payment card at all. A local installation from the open source code, or the Private Server option, leaves the operator in control of everything that installation handles.

How long account data and saved or exported results persist is not stated, and systems that call the API keep their own records under their own practices.

Source: Vendor Published
CC on AI Liability and RecourseLiability is addressed only through a standard limitation clause that disclaims the exposure the product creates.

AI Liability and Recourse

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.

The website terms disclaim warranties about completeness and accuracy and require users to indemnify PQAI against listed claims. The limitation of liability clause still carries placeholders where the cap and lookback period belong, so no figure applies. The API runs under separate terms covering access, use and termination, and the pricing page rules out refunds on every plan. Nothing obliges PQAI to compensate a user for a missed publication or an incomplete result, and the terms of a separately signed commercial order are not public.

Source: Vendor Published
BB on Practice Systems Integration DepthReal integrations exist and are documented, short of depth: named connections without a description of what they actually move.

Practice Systems Integration Depth

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.

The REST API takes a plain language description and returns structured patent results to another application, with documentation covering authentication, endpoints and sample responses, and a patent data endpoint returns claims and bibliographic details. An embeddable interface and a guide to connecting PQAI to an outside AI assistant are also offered. In the technology transfer case, the office attached returned references to its existing disclosure records for staff and outside counsel.

API Starter, Developer and Enterprise plans set the request volume. These suit custom integrations, and no ready made connection syncs patent research with a law firm's matter or document repository.

Source: Vendor Published
AA on Deployment Model and Data ResidencyDeployment options and data residency are published, including the regions available, what changes between tiers, and where processing happens as distinct from where data is stored.

Deployment Model and Data Residency

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.

Users reach the hosted web tool and API or pull the code from GitHub and run their own installation, and a Private Server option is also offered. Hosted requests travel to PQAI over an encrypted connection, while a local installation puts infrastructure, data and access in the operator's hands. Documentation covers the local option and integration. The hosted service names no region and makes no contractual residency commitment, so local installation is the route for teams that need to keep data on their own systems.

Source: Vendor Published
CC on Security Certifications and Trust CenterBadges appear on the site with no scope, no date, and no report available.

Security Certifications and Trust Center

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.

PQAI's security material describes HTTPS, token authentication for API calls, secure infrastructure and access controls, and the hosted terms state that user requests are not stored. Buyers with requirements such as GDPR or SOC 2 are asked to contact the project to discuss them. No independent certification, trust center or audit report covers PQAI itself, and the open source code is the main way to inspect how it handles data.

InspireIP, which manages the application, advertises a SOC 2 examination for its own separate invention disclosure software.

Source: Vendor Published
BB on Model Supply Chain DisclosureThe supply chain is partly disclosed: providers named without change notification, or architecture described without the providers.

Model Supply Chain Disclosure

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.

Project materials describe a purpose built retrieval model trained on historical USPTO examination data, with the retrieval algorithm contributed by GreyB and the open source components released on GitHub. Change notes record updates to the model in production. User requests go to no third party AI engine, so the hosted model is PQAI's own. Guides explain how to call PQAI from outside AI assistants, which run their own models under their own providers.

The hosted service does not list the version and configuration of every deployed model and endpoint, and customers cannot pick a different hosted ranker.

Source: Vendor Published
BB on Commercial TransparencyReal pricing is published for part of the range, with enterprise tiers withheld, or the unit and structure are stated without the figure.

Commercial Transparency

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.

The free plan covers more than 100 million patents with up to 20 results per query, classification prediction, concept extraction and patent data retrieval, with no export or API access. PQAI+ costs $20 a month or $216 a year and raises that to 500 results per query, with full text excerpts, export, 20 personal API requests a month and priority support. Developer API lists at $199 a month and Enterprise at $700, both at half price in a limited time offer with no end date.

Commercial research such as client work, law firm research, invention disclosure and due diligence is allowed on every plan, while building PQAI into a product needs an Enterprise API plan. No plan is refundable, and cancellation keeps access to the end of the period.

Source: Vendor Published
BB on Firm and Practice CoverageSegment and practice coverage is described with substance, short of the boundaries: what is supported is clear, what is not is left open.

Firm and Practice Coverage

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.

Patent attorneys, examiners, inventors, research teams and university technology transfer offices are the intended users. The platform supports early invention screening, patentability work, invalidity research, classification and disclosure review, drawing on worldwide patents and nonpatent publications. Separate technical interfaces let institutions embed the retrieval in existing tools. A law firm can examine ranked publications with their dates, claims and classifications, and client work is a permitted use on every plan. Coverage centers on prior art research rather than litigation document management, docketing or filing.

Source: Vendor Published
Sources on file

5 public documents

The public pages on file for PQAI, 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.

Pricing

$20 per month for PQAI+

USD, as published, never converted
  • Free covers more than 100 million patents with 20 results per query, and commercial research is allowed on every plan.
  • PQAI+ costs $20 monthly or $216 annually, with 500 results per query, export and 20 personal API requests a month.
  • API Starter costs $20 monthly for 50 United States patent lookups. Developer lists at $199 and Enterprise at $700 a month, both half price in a limited time offer.
  • Integrating PQAI into a product requires an Enterprise API plan.

Hosted tiers are individual monthly or annual subscriptions, and API tiers carry their own query allowances and terms. Commercial research is permitted on all plans, and commercial product integration requires an Enterprise API plan.

Implementation: Technical help with API implementation is offered, with no separate implementation price listed.

Note: The 50% offer on the Developer and Enterprise API plans shows no end date. No plan is refundable, and cancellation keeps access to the end of the billing period.

Legal Signals

What each signal means

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.

Confidentiality and Privilege

Client Data in Training

Can material a lawyer puts into this product be used to train a model?

Never, in policy only

A public policy or trust page states no training on customer content, with no matching term located in the published agreement.

The security page says user input is not fed into any training pipeline or third party AI engine, and the website terms state that it is not stored. The models train on USPTO examination data rather than on what users type.

Source: Vendor PublishedYour queries are not fed into any training pipeline or third-party AI engine.As of Nov 19, 2026Evidence

Prompt and Output Retention

How long does the product keep what a lawyer typed, and can that be set to zero?

Disclosed without a period

Retention is acknowledged in public materials with no stated period.

Hosted requests are not stored on PQAI's servers. No single schedule covers saved results, exported files, external connections or account records.

Source: Vendor PublishedAs of Nov 19, 2026Evidence

Ethical Walls and Matter Segregation

Does retrieval respect the firm’s ethical walls, or can the model read across them?

Not addressed

No located public material addresses walls or matter level segregation.

Hosted accounts are individual, and the software can run locally under the operator's own access controls. No ethical wall enforcement or inherited matter permissions are described.

Source: Operator VerifiedAs of Nov 19, 2026Evidence

Third Party Request and Subpoena Notice

If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?

Not addressed

No located term or policy addresses third party requests for customer data.

The public terms make no promise to alert users about compulsory disclosure demands. Because hosted requests are not stored, little of what users type is held to disclose.

Source: Operator VerifiedAs of Nov 19, 2026Evidence
Accuracy and Authority

Primary Law Corpus Provenance

Where does the law in this product come from, and does the vendor have the right to use it?

Sources named, basis unstated

Sources are identified without stating the license or rights basis.

USPTO examination data is the named source of the training sets, and the collection covers more than 100 million patents plus nonpatent literature. The website's data reuse terms name no specific license for the complete collection.

Source: Vendor PublishedAs of Nov 19, 2026Evidence

Good Law Verification

Does the product tell you when the authority it just cited has been overruled?

Not addressed

No located public material addresses whether authority is checked for subsequent history.

The platform retrieves patents and related publications. There is no citator and no check of how courts have treated an authority.

Source: Operator VerifiedAs of Nov 19, 2026Evidence

Refusal and Uncertainty Behavior

What does the product do when the answer is not in the corpus?

Not addressed

No located public material addresses what the product does when it cannot ground an answer.

Results come back as relevance ranked references to inspect. Nothing describes a refusal or a point at which the system declines to suggest a legal conclusion.

Source: Operator VerifiedAs of Nov 19, 2026Evidence

Fabricated Citation Record

Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?

None located

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.

No court order or disciplinary decision in the AI Hallucination Cases database attributes a fabricated citation to PQAI. The platform returns patent publications rather than legal argument.

Source: Operator VerifiedAs of Nov 19, 2026Evidence
Professional Responsibility

Bar Guidance Alignment

Has the vendor engaged in public with the ethics opinions its buyers are bound by?

Not addressed

No located public material engages with bar or ethics guidance.

Patent professionals doing early prior art work are the intended users. No bar opinion or patent office guidance on practitioners' use of AI is mapped to the tool.

Source: Operator VerifiedAs of Nov 19, 2026Evidence

Billing and Fee Posture

Does the vendor address what happens to the bill when the work takes an hour instead of six?

Savings claims only

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.

PQAI points to faster review of invention disclosures, including one office's 30% increase in disclosures reviewed. Nothing covers lawyer timekeeping or disclosing AI related fees to clients.

Source: Vendor PublishedAs of Nov 19, 2026Evidence

Outside Counsel Guideline Readiness

Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?

Not addressed

No located public material supports a client side disclosure obligation.

The security and privacy pages describe how user requests are handled and protected, and compliance requirements such as GDPR or SOC 2 are discussed directly with the project. No outside counsel disclosure package or complete processor list for the web tool and API is available.

Source: Operator VerifiedAs of Nov 19, 2026Evidence

Court Disclosure Support

If a judge’s standing order requires an AI disclosure, can the product produce one?

Not addressed

No located public material addresses court disclosure or verification certification.

PQAI+ users can export results for offline review. Nothing reports prompts, outputs and model use in a form suited to a court disclosure.

Source: Operator VerifiedAs of Nov 19, 2026Evidence
Contact

Correct a record, or ask how something was graded

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.

AI Legal Index

The AI Legal Index is an independent index that tracks changes to AI vendors in legal. It holds 535 vendors across 9 categories, each graded on the same 15 capability axes and recorded against 12 legal signals, from public sources. No vendor pays for inclusion, placement, or rating.

Index Status
Last index update
November 19, 2026
The AI Legal Index is an editorial reference. It is not a regulatory body, not a law firm, and nothing published here is legal advice or a recommendation to retain or avoid a vendor. Records are verified against published sources, bar guidance and public court records. Where a record reads not addressed, the material was not located in public sources on the date shown. See the Methodology page for evaluation standards and limitations.
© 2026 AI Legal Index
3801 N Capital of Texas Hwy, Ste E240 · Austin, TX 78746