Skip to content

Government · Moraph software

Patent & Trademark AI Examiner

Find the prior art and conflicting marks sooner, and leave every determination with the examiner

Overview

The Patent & Trademark AI Examiner searches patent documents and scientific literature for prior art relevant to each claim of an application, and searches registered and pending marks for those that may conflict with a trademark application. It ranks what it finds, shows the passages that match each claim element, and keeps the full search history. Examiners decide what is relevant and make every determination; the tool records the searches behind them.

The problem it solves

Examiners have a limited amount of time for each application, and much of it goes into building search queries across several databases, reading through long result lists and recording what was searched. Relevant prior art may sit in a foreign-language document or a journal article that a keyword search misses, and similar marks may differ in spelling or sound rather than in letters. When a search is questioned later, during quality review, appeal or litigation, the record of where the examiner looked and why can be thin.

Who it’s for

  • Patent examiners and supervisory patent examiners in a national intellectual property office
  • Trademark examining attorneys and their managers
  • Search and classification specialists who support examination
  • Quality assurance staff who review the completeness of searches
  • Office leadership and oversight bodies who need to see how AI is used in examination

How the work flows

  1. Understand the applicationThe tool reads the claims, specification and classification of a patent application, or the mark, goods and services of a trademark application, and breaks them into the elements a search needs to cover.
  2. SearchEach claim element or mark is searched across the patent, literature and trademark sources your office makes available, using meaning as well as keywords, and across languages where the source supports it.
  3. ReviewThe examiner reviews ranked results with the matching passages highlighted against each claim element, marks what is relevant and why, and sets aside the rest.
  4. RecordThe examiner's selected references, claim chart and search history are exported to the office's case system to support the action the examiner writes. The tool does not write or issue office actions.

What it does

Claim element breakdown
Claims are split into elements so each one can be searched and charted separately; the examiner can edit the breakdown.
Meaning-based search
Documents are found by what they describe as well as the words they use, alongside conventional keyword and classification searches.
Cross-language search
Documents in other languages are searched and shown with machine translations, clearly marked, for the examiner to check against the original.
Non-patent literature
Scientific and technical literature from sources your office licenses is searched in the same step as patent documents.
Highlighted passages
Each result shows the passages that match each claim element, so the examiner can judge relevance without reading the whole document first.
Claim charts
Passages the examiner accepts are assembled into a claim chart by element, ready to export to the case system.
Similar-mark search
Trademark applications are searched for registered and pending marks that look, sound or mean alike, with overlapping goods and services shown.
Design element search
Marks with design elements are searched by the design codes your office uses, alongside the words in the mark.
Complete search history
Every query, source, date and result the examiner reviewed is kept as the record of the search.
Examiner-controlled relevance
Only the examiner marks documents as relevant; the tool's ranking is a suggestion and is labeled as one.
Quality review view
Reviewers see the search history and selected references beside the action that followed, and record their findings.
AI use record
Each application records which AI functions were used, which model version produced the results and what the examiner accepted or rejected.

Who uses it

Patent examiner
Searches for prior art, reviews results, builds claim charts and records the search
Trademark examining attorney
Searches for conflicting marks and records the search behind each determination
Supervisory examiner
Oversees examiners' work, reviews searches and manages assignments
Search specialist
Runs supporting searches on request and maintains search configurations
Quality reviewer
Reviews samples of searches and records findings
Auditor (read-only)
Views search records, AI use records and the audit trail, with no ability to change them

Integrations

Case management system
Applications, claims, classifications and assignments from your office's examination case system, by API or scheduled transfer.
Patent document collections
Published patent documents from the collections your office holds or receives under its data exchange arrangements, indexed during implementation.
Non-patent literature sources
Scientific and technical literature from the databases your office licenses, searched under the terms of those licenses.
Trademark register
Registered and pending marks, goods and services, and design codes from your office's trademark records.
Single sign-on
Staff sign in through your identity provider (SAML or OpenID Connect), including PIV-backed authentication where your office uses it.
Work API
Search activity, AI use records and quality review results available to your data warehouse and reporting tools through the Moraph Work API.

Security

Confidential applications
Unpublished application content is visible only to the people assigned to it and their supervisors, supporting the confidentiality the law requires.
No training on your applications
Application content and examiner decisions are not used to train models shared with other clients.
United States hosting
Data is hosted in Moraph's United States region only, and support tickets from federal clients are handled only by US citizens with active security clearances.
Encryption and keys
Client data is encrypted, with keys Moraph staff cannot use without explicit, time-bound authorization.
Complete audit trail
Every search, view, relevance decision and export is logged with the person and time, and cannot be edited.
Tested releases
Every release passes security testing gates before it ships, and a software bill of materials is maintained for the product. Moraph is SOC 2 Type II and ISO/IEC 27001 audited.

Hosting and deployment

Moraph cloud, United States
Hosted and operated by Moraph in the United States region, with data kept in the US.
Dedicated environment
A single-tenant environment for offices that require isolation, including isolated model hosting, scoped in the proposal.
With a managed service
Optionally run alongside a Moraph managed service for administration, source updates and integration support.

Implementation

Source onboarding
We index the patent collections, literature sources and trademark records your office makes available, within the terms of each source's license.
Case system integration
Applications, assignments and exports are connected to your examination case system. We need interface documentation and a contact from the system's owners.
Evaluation against past cases
The search is run on a set of already-examined applications and the results compared with the references examiners cited, so your office can judge its usefulness before go-live.
Roles and workflow
Roles, art units or law offices, quality review sampling and export formats are configured to match your procedures.
AI governance documentation
We document the models used, their intended use, their known limitations and the human review steps, for your office's AI inventory and governance process.
Pilot and training
A pilot group of examiners uses the tool in selected areas for a period you choose, and examiners, supervisors and reviewers are trained on their own workflows.

What it doesn’t do

  • It does not decide patentability or registrability. Examiners decide what is relevant and make every determination, and the office's procedures apply.
  • It can miss relevant prior art or conflicting marks. Its results supplement, and do not replace, the examiner's own search.
  • It searches only the sources your office makes available and licenses. Coverage of any source depends on what that source contains.
  • Machine translations can be wrong; examiners should check relevant passages against the original or an official translation.
  • It does not write or issue office actions; it exports references, claim charts and search history to your case system.
  • Moraph holds no FedRAMP authorization for this product. We provide control documentation to support your office's own authorization decision.
  • It is not designed for classified information, including applications under secrecy orders, which must be handled outside it.
  • Ranking quality varies by technology area and should be reviewed periodically with quality assurance staff.

How to evaluate it

  • Run the search on already-examined applications and compare its results with the references examiners cited
  • Ask examiners from different technology areas to work an application and note where results helped or misled
  • Check machine-translated passages against originals for a sample of foreign-language documents
  • Test similar-mark search on marks that differ by sound or spelling rather than letters
  • Review the exported search history and claim chart against what your quality reviewers need
  • Confirm that unpublished applications are visible only to the people assigned to them

Questions

Does the tool decide whether a claim is patentable or a mark can be registered?

No. It finds and ranks documents and marks. Examiners decide what is relevant and make every determination under your office's procedures.

Can examiners still run their own searches?

Yes. Examiners add their own keyword, classification and other queries, which run alongside the tool's and are kept in the same search history.

Are our applications used to train models for others?

No. Application content and examiner decisions are not used to train models shared with other clients.

Which databases does it search?

The patent collections, literature sources and trademark records your office makes available and licenses. Sources are agreed and indexed during implementation.

Is it FedRAMP authorized?

No. Moraph holds no FedRAMP authorization for this product. We provide control documentation to support your office's own authorization decision.

How do we judge whether it helps?

During implementation the search is run on applications your examiners have already examined, and the results are compared with the references they cited, so your office can judge it on its own cases.