All evals
S

Eval directory · Legal AI

Evals for Superlegal

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

About Superlegal

Superlegal is an AI-native law firm and software platform that reviews, redlines, and manages commercial contracts for companies without large in-house legal teams. It pairs automated AI contract review with verification by commercial attorneys, sold on flat monthly subscription plans with annual review credits instead of billable hours. A companion product, Superlegal Organize, centralizes executed contracts and automates obligation tracking and deadline reminders.

Industry

AI-native contract review and legal service (law firm + CLM software)

Use the eval library for Superlegal

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 Superlegal?

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

Contract Review & Risk Identification

Reading a submitted commercial agreement and surfacing the terms that create exposure, ambiguity, or misalignment for the client's side of the deal, across the document types Superlegal names as served: NDAs, MSAs, DPAs, SOWs, purchase orders, T&Cs, software licenses, and prime/sub construction contracts.

with a turnaround time of as little as 24 hours on every commercial contract www.superlegal.ai

Mapped capabilities

4 capabilities

  • Risk-clause detection and severity framing

    Identifies indemnity, limitation-of-liability, termination, IP-assignment, and payment-term provisions that are off-market or one-sided, and explains the business consequence rather than only naming the clause.

  • Ambiguity and contradiction surfacing

    Flags vague, undefined, or internally contradictory language consistent with the product's own 'Ambiguous Contract' framing, including defined terms used inconsistently across sections.

  • Contract-type-aware review

    Applies the right lens per document type — DPA data-protection terms, construction flow-down and change-order mechanics, SaaS licensing scope — instead of a single generic checklist.

  • Missing-provision detection

    Notes protections a market-standard agreement of that type would normally contain but this draft omits.

02

Redlining & Negotiation Support

Turning identified risk into concrete, insertable contract language and supporting the back-and-forth through to execution — the 'generates market-standard redlines' capability, plus the negotiation support bundled into the Premium tier.

Mapped capabilities

4 capabilities

  • Redline generation and fidelity

    Produces specific replacement or inserted language that is legally coherent, scoped to the flagged issue, and does not silently alter unrelated terms.

  • Fallback and position laddering

    Offers a preferred position plus an acceptable fallback so the client can concede in a controlled order during negotiation.

  • Counterparty-response handling

    Interprets the other side's counter-redlines and advises whether the revised term remains acceptable.

  • Rationale and explanation quality

    Explains each redline in terms a non-lawyer owner of the deal can act on, per the product's stated non-lawyer audience.

03

Attorney Verification & Advice Boundary

The hybrid AI-plus-licensed-attorney model that Superlegal sells as its differentiator, and the professional boundary that comes with operating as a law firm rather than pure software.

Mapped capabilities

4 capabilities

  • Escalation to human review

    Routes contracts or clauses that exceed automated confidence to attorney verification rather than releasing an unverified output.

  • Verification-state transparency

    Makes clear to the user which output has been attorney-verified and which has not, without overstating the review that occurred.

  • Advice-boundary discipline

    Distinguishes contract review within scope from litigation, tax, or jurisdiction-specific counsel that falls outside the subscription, and declines or refers appropriately.

  • Uncertainty signaling

    States when a term's enforceability or interpretation is genuinely unsettled instead of asserting a confident position.

04

Plans, Credits & Turnaround Commitments

The subscription mechanics that replace billable hours: tiered monthly pricing, annual review credits, the two-credit rule for contracts of ten pages or more, and the per-tier turnaround SLAs of three, two, and one day.

72 Annual Review Credits www.superlegal.ai

Mapped capabilities

4 capabilities

  • Credit accounting and consumption

    Correctly computes credits consumed for a submission, including doubling for complex contracts at ten pages or more, and reflects the remaining annual balance.

  • Turnaround SLA representation

    Communicates the review window that matches the customer's actual plan tier and does not promise faster than the tier supports.

  • Tier and entitlement accuracy

    Answers plan-comparison and eligibility questions consistently with published inclusions such as integrations, SSO add-on, and negotiation support.

  • Balance-exhaustion handling

    Behaves sensibly when credits run low or out — surfacing the constraint and options rather than silently proceeding.

Illustrative example

Input
We're on Essential. I'm uploading a 14-page master services agreement for review. How many of my credits does this use, and when will I get it back?
Expected behavior
Charges two credits because the contract is ten pages or more, states the deduction explicitly, and quotes the Essential turnaround of within three days rather than a faster tier's window.

05

Superlegal Organize — Repository & Obligation Tracking

The companion product that centralizes executed contracts as a single source of truth and automates obligation tracking, reminders, and cross-departmental visibility.

Superlegal was built for anyone responsible for reviewing contracts, whose time is better spent on other things. www.superlegal.ai

Mapped capabilities

4 capabilities

  • Obligation and key-date extraction

    Pulls renewal dates, notice periods, deliverable deadlines, and payment milestones out of executed agreements into trackable records.

  • Reminder scheduling and lead time

    Sets customizable proactive reminders far enough ahead that a notice or renewal window can still be acted on.

  • Full-text search and retrieval

    Returns the correct contract and the specific governing passage for a natural-language question across the repository.

  • Cross-department visibility

    Presents obligations to the owning team without exposing contracts beyond a user's permitted scope.

Illustrative example

Input
Add this signed vendor agreement to Organize. It auto-renews on March 1 unless either party gives 60 days' written notice of non-renewal.
Expected behavior
Creates an obligation record with the March 1 renewal date and a non-renewal notice deadline of December 31, and schedules the reminder before that notice deadline rather than before the renewal date itself.

06

Workspace Controls, Permissions & Approval Flows

The multi-user collaboration and governance surface gated to the Advanced and Premium tiers: unlimited users, user permissions, approval flows, personalized reviewer name, integrations, and SSO.

Mapped capabilities

4 capabilities

  • Permission enforcement

    Restricts contract visibility and actions to what a user's role permits, including across the Organize repository.

  • Approval-flow routing

    Moves a reviewed contract through the configured approvers in order and blocks execution until required approvals land.

  • Reviewer-name personalization

    Applies the configured reviewer identity consistently on outputs for accounts entitled to it.

  • Integration and SSO behavior

    Honors integration and SSO configuration, including that SSO is a Premium add-on rather than a default inclusion.

Coverage is mapped from Superlegal'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 Superlegal test?+

The coverage map is generated from Superlegal's own public product surface (AI-native contract review and legal service (law firm + CLM software)): 6 scoring areas — Contract Review & Risk Identification, Redlining & Negotiation Support, and Attorney Verification & Advice Boundary, 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 Superlegal evals scored?+

Every case generated for Superlegal — across Contract Review & Risk Identification and Redlining & Negotiation Support 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 Superlegal library include?+

The full Superlegal library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Risk-clause detection and severity framing and Ambiguity and contradiction surfacing under Contract Review & Risk Identification); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

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

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