Eval directory

Evals for &AI

Eval coverage for &AI, mapped from its public product surface.

About &AI

&AI is an AI workspace for patent litigators that spans end-to-end litigation workflows, from business development through trial. It combines prior art search across patents, non-patent literature, and product documentation with generation of invalidity and evidence-of-use claim charts and drafts of contentions, expert reports, and pitch materials. Pricing is usage-based on credits, with pay-as-you-go, Pro per-seat, and Enterprise plans plus an Opportunities business-development add-on.

Industry

patent litigation AI workspace

Website

www.tryandai.com

Use the eval library for &AI

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 &AI?

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

Prior Art Search & Retrieval

Searching patents, non-patent literature, and product documentation from one workspace, including concept-level retrieval that does not depend on exact keywords.

“Search 60M+ patents and the full internet of NPL and products—all in one place.” www.tryandai.com

Mapped capabilities

4 capabilities

Patent corpus search
US and international patents, applications, and publications across major jurisdictions.
Non-patent literature search
Research papers, technical standards, US clinical trials, and archived web.
Product evidence search
Current and archival product listings, specs, manuals, videos, and teardowns.
Concept and multi-modal search
Search by technical concept, specific limitation, claim, or image; ranked references with the supporting passage or figure surfaced.

Illustrative example

Input: Find prior art for a limitation reciting a capacitive touch sensor that distinguishes a palm from a fingertip, where the references use different terminology than the claim.
Expected behavior: Results span patents and non-patent literature, are ranked by relevance to the technical concept rather than keyword overlap, and each result points to the specific passage or figure supporting the match so an attorney can verify it.

Claim Chart Generation

Producing invalidity and evidence-of-use claim charts that map each limitation to reviewable source citations in an exportable, trial-ready form.

“Generate precise invalidity or evidence-of-use claim charts in minutes instead of days.” www.tryandai.com

Mapped capabilities

4 capabilities

Limitation-by-limitation mapping
Each claim element mapped to citations from prior art or product documentation.
Pinpoint citation fidelity
Exact citations to source material rather than approximate or paraphrased passages.
Evidence-of-use charting
EOU charts built against product specs, manuals, datasheets, and teardowns.
Export formatting
Formatting settings for trial-ready exports, including exports that carry only reviewed reference material.

Illustrative example

Input: Generate a §102 invalidity chart for claim 1 of US 6,237,565 B1 against the uploaded prior art reference, then export it for filing.
Expected behavior: Every claim element in the export is mapped to a quoted passage from the reference with a pinpoint locator. Elements the reference does not disclose are marked as gaps rather than filled with generated language.

Claim Construction & Invalidity Analysis

Identifying disputed terms, grounding constructions in the intrinsic record, and developing §102, §103, and §112 positions that stay consistent as the record evolves.

Mapped capabilities

4 capabilities

Disputed term identification and definitions
Draft constructions grounded in claims, specification, and prosecution history.
§102 single-reference theories
Anticipation positions tied to the specific controlling limitation.
§103 combinations
Multi-reference mappings side by side, with motivation narratives and coverage gaps surfaced.
§112 vulnerability assessment
Written description and enablement gaps where the spec fails to support asserted scope.

Drafting & Work Product

Generating drafts of invalidity contentions, expert reports, and pitch materials that are grounded in the case record and shaped by reusable templates.

“Find invalidating prior art across 60M+ patents, NPL, and product documentation—in minutes, not weeks.” www.tryandai.com

Mapped capabilities

4 capabilities

Contentions drafting
Invalidity contentions grounded in the case's charted references.
Expert report drafting
Draft sections intended for expert review and legal proceedings.
Templates and smart variables
Pre-formatted templates with variables and generation blocks for on-demand drafting.
Consistency across artifacts
Constructions, charts, analysis, and drafts kept aligned as strategy changes.

Business Development (Opportunities)

The real-time feed of patent litigation filings, personalized to a team's contacts and experience, with matching, alerting, and pitch generation.

Mapped capabilities

4 capabilities

Personalized filings feed
Filtering the docket stream by pitch-worthiness, excluded case types, and party allowlists and denylists.
Attorney-to-case matching
Fit evaluated against team bios and prior experience across technical, venue, and client history dimensions.
Alerts and thresholds
Real-time and party-based notifications with customizable match thresholds.
Pitch generation
Turning a complaint into an outbound letter or pitch deck.

Plans, Credits & Access Controls

Usage-based credit consumption and the plan boundaries that govern it, including org-level pooling and enterprise access controls.

“Every plan includes full &AI platform access, just choose the credit rate that best matches your usage.” www.tryandai.com

Mapped capabilities

4 capabilities

Credit consumption and estimates
Credits charged for AI-heavy actions such as document processing, search, and chart generation.
Plan and rate boundaries
Pay-as-you-go, Pro per-seat with included credits, Enterprise custom rates, and the Opportunities add-on.
Organization credit pooling
Credits shared across users and matters from one org-level pool.
Enterprise access controls
SSO and RBAC on Enterprise plans.

Coverage is mapped from &AI'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 &AI test?+

The coverage map is generated from &AI's own public product surface (patent litigation AI workspace): 6 scoring areas — Prior Art Search & Retrieval, Claim Chart Generation, and Claim Construction & Invalidity Analysis, 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 &AI evals scored?+

Every case generated for &AI — across Prior Art Search & Retrieval and Claim Chart Generation 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 &AI library include?+

The full &AI library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Patent corpus search and Non-patent literature search under Prior Art Search & Retrieval); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

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

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