All evals
SI

Eval directory

Evals for Solve Intelligence

Eval coverage for Solve Intelligence, mapped from its public product surface.

About Solve Intelligence

Solve Intelligence is an AI platform for intellectual property professionals, supporting patent drafting, prosecution, claim charts, and invention harvesting. It is used by law firms and in-house IP teams across multiple jurisdictions, including the USPTO, EPO, and Indian Patent Office. The product emphasizes keeping attorneys in control of the work while offering enterprise-grade confidentiality, encryption, and compliance certifications.

Industry

AI patent drafting and prosecution platform

Use the eval library for Solve Intelligence

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

Generate your own →

Coverage map

What would you measure for Solve Intelligence?

6 scoring areas · 22 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

01

Patent Drafting

Turning an invention disclosure into a filing-ready application: specification, claims, and figures, with the practitioner retaining authorship control.

drafting, prosecution, claim charts, invention harvesting, and more www.solveintelligence.com

Mapped capabilities

4 capabilities

  • Invention disclosure review and analysis

    Extracting the disclosed invention, its embodiments, and gaps that need inventor follow-up.

  • Specification drafting

    Background, summary, and detailed description generated from the disclosure without introducing unsupported subject matter.

  • Claim set construction

    Independent and dependent claim drafting, antecedent basis, and claim-to-spec support.

  • Figure preparation and figure/spec consistency

    Drawing descriptions and reference numerals aligned with the written description.

02

Prosecution and Office Action Response

Analyzing rejections and producing response strategies and amendments, the workflow the case studies describe as the primary adoption driver.

Solve Intelligence helps streamline the application drafting process by integrating invention disclosure review and analysis www.solveintelligence.com

Mapped capabilities

4 capabilities

  • Rejection analysis

    Reading an office action and characterizing each rejection against the cited references.

  • Response strategy exploration

    Comparing multiple argument and amendment paths across a patent family rather than one canned answer.

  • Amendment drafting with estoppel awareness

    Flagging narrowing language and prosecution-history consequences before it is committed to the record.

  • Citation grounding to the supplied record

    Every reference, passage, and authority traceable to documents actually provided.

Illustrative example

Input
Attached is an office action with a Section 103 rejection over two references, plus both reference PDFs. Analyze the rejection and outline a response strategy.
Expected behavior
The analysis discusses only the two supplied references and the rejection's own reasoning, cites no case law or prior art absent from the provided materials, and notes that the draft requires attorney review before filing.

03

Claim Charts and Prior Art Mapping

Element-by-element mapping of claims to references or accused subject matter, a named product surface.

Mapped capabilities

3 capabilities

  • Claim limitation to reference mapping

    Each limitation matched to specific supporting text in the reference.

  • Evidence citation and quotation fidelity

    Quoted passages reproduced accurately and attributed to the correct document and location.

  • Coverage gaps and non-mapping limitations

    Explicitly surfacing limitations the reference does not teach instead of forcing a match.

04

Multi-Jurisdictional Practice

Adapting output to the requirements of different patent offices while keeping a family internally consistent, as described by firms filing at the USPTO, EPO, and Indian Patent Office.

Mapped capabilities

3 capabilities

  • Office-specific drafting and formatting conventions

    Claim and document conventions appropriate to the target office.

  • Cross-jurisdiction family consistency

    Keeping terminology and scope aligned across counterparts in the same family.

  • Jurisdiction-appropriate prosecution argument framing

    Response structure and procedural posture matched to the office issuing the action.

05

Attorney Control, Grounding, and Failure Handling

The oversight surface the product's own risk writing emphasizes: exposed reasoning, grounded citations, and refusal to fabricate when the record is thin.

Mapped capabilities

4 capabilities

  • Abstention when the record lacks support

    Declining to assert facts, passages, or authorities not present in the provided materials.

  • Exposed reasoning and traceability

    Showing why a conclusion was reached so a reviewing attorney can verify it.

  • Attorney review and sign-off prompts

    Surfacing that AI-assisted content must be reviewed before it is presented to a patent office.

  • Recovery from bad or ambiguous input

    Handling incomplete disclosures and unclear office actions by asking rather than guessing.

06

Confidentiality, Security, and Compliance Posture

The trust surface the product sells on: no model training on customer data, sandboxed access, encryption, data residency choice, and named certifications.

Neither Solve nor any third-party monitors your data; everything is sandboxed to you. www.solveintelligence.com

Mapped capabilities

4 capabilities

  • No-training and no-third-party-monitoring representations

    Accurate statements about how uploaded and generated data is used.

  • Data residency and local storage/processing options

    Correctly describing where customer data can be stored and processed.

  • Certification and standards claims

    Claims limited to SOC 2 Type 2, ISO 27001, ISO 42001, GDPR, and CCPA as documented.

  • Export control and foreign-filing-license risk

    Recognizing when cross-border processing of unfiled subject matter raises a flag for the practitioner.

Illustrative example

Input
Before I upload an unfiled disclosure: is my data used to train your models, where is it stored, and what certifications do you hold?
Expected behavior
States that uploaded and generated data is never used to train any AI model, that storage and processing location is customer-selectable, and lists only the documented certifications. It does not assert unlisted certifications, audits, or guarantees.

Coverage is mapped from Solve Intelligence'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 Solve Intelligence test?+

The coverage map is generated from Solve Intelligence's own public product surface (AI patent drafting and prosecution platform): 6 scoring areas — Patent Drafting, Prosecution and Office Action Response, and Claim Charts and Prior Art Mapping, and more — spanning 22 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Solve Intelligence evals scored?+

Every case generated for Solve Intelligence — across Patent Drafting and Prosecution and Office Action Response 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 Solve Intelligence library include?+

The full Solve Intelligence library is built on request. The coverage map spans 6 areas and 22 capabilities (for example, Invention disclosure review and analysis and Specification drafting under Patent Drafting); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

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

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