Eval directory

Evals for Patlytics

Eval coverage for Patlytics, mapped from its public product surface.

About Patlytics

Patlytics is an AI-native patent platform for IP teams, covering infringement detection, invalidity, freedom-to-operate, SEP, portfolio analysis, patent and NPL search, and claim construction in one product. It produces citation-backed claim charts and prior art analysis by crawling public sources and searching large patent and non-patent literature corpora. The company markets to Fortune 500 companies and Am Law 100 firms and recently announced a $40 million Series B.

Industry

AI patent intelligence platform for IP professionals

Website

www.patlytics.ai

Use the eval library for Patlytics

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Patlytics?

6 scoring areas · 24 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Infringement Detection & Claim Charting

Evidence-of-use workflows that crawl public sources or ingest customer-supplied materials and produce citation-backed claim charts, at single-patent and portfolio scale.

“Delivering the most accurate data on the market, with citation-backed outputs, source auditing, and color-coded confidence indicators.” www.patlytics.ai

Mapped capabilities

4 capabilities

Evidence-of-use crawling across public sources
Locating teardowns, datasheets, filings, and product pages as candidate evidence for a target product.
Citation-backed claim chart construction
Mapping each claim limitation to a located source with a resolvable pin cite, including unmatched limitations.
Auto-generated claim construction as chart input
Deriving limitation-level constructions and carrying them consistently into the chart.
Bring-your-own evidence ingestion
Charting strictly against uploaded documents when the user supplies their own evidence set.

Illustrative example

Input: Chart claim 1 of US 9,876,543 against the Acme X200 router using only my two uploads: a teardown PDF and the product datasheet. Do not crawl.
Expected behavior: Each limitation row cites a specific page or section of one of the two uploaded documents. Limitations with no support in those documents are marked unmatched or for review rather than filled with plausible-sounding text, and no outside source is introduced.

Invalidity & Prior Art Analysis

Novelty and inventive-step analysis over patent and non-patent literature, producing contention-style charts and obviousness combinations with stated rationale.

“Obviousness combinations with motivations to combine, plus written description & enablement” www.patlytics.ai

Mapped capabilities

4 capabilities

Novelty (§102) mapping to a single reference
Anticipation analysis with per-limitation disclosure and pin cites.
Obviousness (§103) combinations with motivation to combine
Multi-reference combinations that state why a skilled artisan would combine them.
Contention-style chart formatting
Multi-reference chart layout where every asserted cell is cited to its source.
Written description and enablement (§112) review
Support and enablement analysis against the specification.

Illustrative example

Input: Build an obviousness contention chart for claim 1 of US 8,123,456 combining US '231 and EP '808.
Expected behavior: The chart pin-cites each asserted disclosure to a paragraph, column, or figure of the specific reference that discloses it, and states an explicit motivation to combine the two references. Limitations disclosed by neither reference are left blank.

Freedom to Operate & SEP Workflows

Clearance analysis starting from a user's own product materials, and standard-essential patent analysis as a distinct announced workflow.

Mapped capabilities

4 capabilities

FTO intake from product materials
Starting a clearance search from uploaded descriptions of an unlaunched product.
Risk-relevant patent identification and triage
Surfacing and ranking patents implicated by the described product.
SEP analysis workflow
Essentiality-oriented analysis against standards, per the announced next-generation SEP workflow.
Scoping and jurisdiction handling
Respecting user-specified corpus, family, or jurisdiction limits in clearance work.

Search & Portfolio Intelligence

Patent and non-patent literature search across the advertised corpora, plus portfolio-scale analysis for competitor monitoring and licensing strategy.

“Crawl for evidence of use across public sources — for one patent or hundreds” www.patlytics.ai

Mapped capabilities

4 capabilities

Semantic patent and NPL search
Query interpretation and retrieval across the advertised patent and literature corpora.
Portfolio analysis and heatmaps
Competitor portfolio views used to triage infringement or licensing targets.
Licensing target identification
Combining infringement and validity signals to surface candidate licensees.
Portfolio-scale batch runs
Running the same analysis across hundreds of patents without per-item drift.

Citation Integrity & Trust Controls

The accuracy and safety guarantees the product markets: cited outputs, source auditing, confidence signaling, hallucination safeguards, and customer-data handling commitments.

“Customer data is encrypted in transit and at rest, segregated, and never used to train models.” www.patlytics.ai

Mapped capabilities

4 capabilities

Pin-cite resolvability and source auditing
Every asserted disclosure traceable to a locatable passage in a named source.
Confidence indication and review flagging
Marking weak or ambiguous mappings for human review rather than asserting a match.
Abstention on unsupported claims
Declining to fill limitations that no retrieved source supports.
Data segregation and no-train handling
Behavior consistent with the stated encryption, segregation, and no-training-on-customer-data commitments.

Prosecution & Document Workspace

Drafting-side and file-management surfaces named across the demo, news, and blog pages, including life sciences additions.

Mapped capabilities

4 capabilities

Patent drafting and invention disclosure forms
Generating draft material from disclosure inputs.
Office action support
Analyzing rejections and assembling responsive material.
My Drive centralized file hub
Storing, deduplicating, and reusing uploaded documents across workflows.
Life sciences artifacts
Sequence listing management and chemical structure visualization.

Coverage is mapped from Patlytics's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Patlytics test?+

The coverage map is generated from Patlytics's own public product surface (AI patent intelligence platform for IP professionals): 6 scoring areas — Infringement Detection & Claim Charting, Invalidity & Prior Art Analysis, and Freedom to Operate & SEP Workflows, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Patlytics evals scored?+

Every case generated for Patlytics — across Infringement Detection & Claim Charting and Invalidity & Prior Art Analysis and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Patlytics library include?+

The full Patlytics library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Evidence-of-use crawling across public sources and Citation-backed claim chart construction under Infringement Detection & Claim Charting); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Patlytics or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Patlytics areas and set them up in a Corsac workspace, where you can run every test case against Patlytics or your own agent with your own data.