All evals
L

Eval directory · Legal AI

Evals for LegalOn

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

About LegalOn

LegalOn is an AI productivity platform for in-house legal teams that covers AI contract review and redlining, drafting, matter intake and management, and contract intelligence across a company's repository. It ships attorney-built playbooks and workflows so teams can apply their own review standards without prompt engineering. Its Vault product turns executed contracts into searchable, structured business intelligence for renewal risk, obligations, and key terms.

Industry

legal AI contract review and matter management platform

Headquarters

San Francisco

Use the eval library for LegalOn

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 LegalOn?

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

01

AI Contract Review & Redlining

Reading a contract in context, surfacing risk, and producing applicable edits rather than flat checklist output. Includes position-awareness (buyer vs. seller side) and one-click application of proposed language.

Unlimited AI contract review and redlining www.legalontech.com

Mapped capabilities

4 capabilities

  • Risk issue spotting in context

    Identifies missing, one-sided, or unusual terms with reference to the clause's role in the agreement; avoids flagging conforming clauses.

  • Position-aware redline generation

    Proposes revised language appropriate to the party the reviewer represents, not a generic neutral rewrite.

  • Edit application and document hygiene

    Applied edits preserve clause numbering, cross-references, and defined-term usage after insertion or deletion.

  • Word-integrated review flow

    Review and edit application operate on the document inside the Word workflow described in the platform surface.

02

Playbooks & Team Standards

Applying a team's own review standards without prompt engineering, using pre-built or customer-configured playbooks so output is consistent regardless of who runs the review.

100+ Attorney-Built Workflows, No Prompt Engineering Required www.legalontech.com

Mapped capabilities

4 capabilities

  • Deviation detection against a playbook rule

    Correctly classifies a clause as conforming, deviating, or missing relative to a specific stated standard.

  • Fallback and escalation positions

    Distinguishes preferred from acceptable fallback language and escalates when a term falls outside both.

  • Reviewer-to-reviewer consistency

    The same contract and playbook produce the same set of flagged deviations across repeated runs.

  • Out-of-the-box playbook coverage

    Pre-built playbooks for common agreement types apply without customer configuration.

Illustrative example

Input
MSA §9 caps liability at fees paid in the prior 3 months. The company playbook requires a 12-month fee cap. Review §9 on behalf of the customer.
Expected behavior
Flags §9 as a playbook deviation, names the 12-month fee-cap rule as the governing standard, and proposes a redline raising the cap to 12 months of fees. Conforming clauses are not flagged.

03

Drafting & Clause Generation

Producing new contract language and full drafts grounded in attorney-built content and templates, including rewriting existing provisions on request.

Mapped capabilities

4 capabilities

  • Clause drafting to a stated standard

    Generates language that satisfies the requested protection or allocation without introducing unrequested terms.

  • Rewriting existing provisions

    Rewrites a supplied clause to change a specific term while leaving unrelated obligations intact.

  • Template-grounded drafting

    Drafts start from the applicable attorney-built template for the agreement type rather than free-form generation.

  • Definition and internal consistency

    New language reuses the document's existing defined terms instead of inventing parallel definitions.

05

Vault: Contract Portfolio Intelligence

Turning executed contracts into searchable, structured intelligence across the repository — renewal risk, obligations, and key terms — for both legal and business consumers.

Vault turns executed contracts into searchable, structured business intelligence www.legalontech.com

Mapped capabilities

4 capabilities

  • Key-term extraction into structured fields

    Extracts terms such as term length, renewal mechanics, and notice periods into consistent, typed fields.

  • Renewal and expiry risk surfacing

    Correctly identifies which agreements renew or lapse within a stated window, including notice deadlines.

  • Portfolio-level filtered queries

    Multi-condition queries across the repository return the complete matching set with no out-of-scope contracts.

  • Provenance and citation

    Every extracted value or query result traces back to the specific source contract and clause.

Illustrative example

Input
Across the contract repository: which vendor agreements auto-renew within the next 90 days and require more than 30 days' notice to terminate?
Expected behavior
Returns only agreements meeting both conditions, listing the renewal date and notice period for each with a citation to the source contract. Agreements outside the window or without auto-renewal are excluded.

06

Matter Intake & Management

Capturing inbound legal requests, routing and tracking them, and reusing prior matters so the team's first response is fast and not dependent on one person's memory.

Mapped capabilities

4 capabilities

  • Request capture and structuring

    Converts an inbound request into a matter with the requester, agreement type, counterparty, and asked-for outcome.

  • Matter summarization

    Produces a summary a second attorney could pick the matter up from without rereading the thread.

  • Prior-matter recommendation

    Surfaces relevant past matters for a new request and explains the basis for the match.

  • Status and ownership tracking

    Matter state and assignee remain accurate as the request moves from intake through completion.

Coverage is mapped from LegalOn'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 LegalOn test?+

The coverage map is generated from LegalOn's own public product surface (legal AI contract review and matter management platform): 6 scoring areas — AI Contract Review & Redlining, Playbooks & Team Standards, and Drafting & Clause Generation, 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 LegalOn evals scored?+

Every case generated for LegalOn — across AI Contract Review & Redlining and Playbooks & Team Standards 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 LegalOn library include?+

The full LegalOn library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Risk issue spotting in context and Position-aware redline generation under AI Contract Review & Redlining); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

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

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