Eval directory

Evals for Marveri

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

About Marveri

Marveri is an AI platform for corporate transactions that ingests a data room and turns unstructured deal documents into organized, reviewable output for legal and deal teams. It targets M&A diligence (buy-side and sell-side), venture financings, and ongoing corporate housekeeping, producing diligence reports, request lists, disclosure schedules, file organization, cap table tie-outs, and missing-document audits. It is sold to law firms, investment banks, PE/VC firms, and corporate development teams, with security positioning around SOC 2, GDPR, encryption, and no training on customer data.

Industry

AI due diligence platform for M&A and venture financings (legal tech)

Website

www.marveri.com

Use the eval library for Marveri

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 Marveri?

6 scoring areas · 23 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

Data Room Ingestion & Organization

Taking an unstructured, inconsistently named set of deal files and producing an organized, navigable data room that follows a firm's conventions.

“Automated file organization, cap table tie-outs, document comparisons, and missing document audits.” www.marveri.com

Mapped capabilities

4 capabilities

Document type classification
Assigning each file to the right category (organizational documents, board and stockholder consents, material contracts, employment, financials, tax).
Naming convention normalization
Rewriting inconsistent filenames into a standard entity / document type / counterparty / date format.
Folder structure construction
Placing classified documents into a nested folder hierarchy that matches the requested diligence index.
Messy input handling
Working through duplicates, drafts vs. executed copies, scanned files, archives, and ambiguous filenames without dropping documents.

Contract Review & Clause Analysis

Reading commercial and corporate agreements to surface the provisions that drive transaction risk, at either survey depth or clause-level depth.

Mapped capabilities

4 capabilities

Assignment and change-of-control analysis
Identifying consent requirements and whether they are triggered by a stock sale, asset sale, or merger.
Termination and renewal terms
Extracting term length, auto-renewal, notice periods, and termination-for-convenience rights.
Wide survey across a contract set
Summarizing the same major provisions consistently across hundreds of agreements for comparison.
Targeted deep analysis via chat
Answering a narrow, custom question about specific contracts or clauses rather than running a preset review.

Illustrative example

Input: Our buyer is acquiring 100% of the target's stock. Does the Meridian Health Systems MSA require Meridian's consent for that deal? Cite the clause.
Expected behavior: Answers whether consent is required based on the MSA's assignment and change-of-control language, distinguishing a stock sale from an assignment of the agreement, and cites the governing section in the named file. Does not generalize from other contracts in the data room.

Diligence Report & Issue Flagging

Turning the reviewed corpus into a diligence report that a deal team can act on before the first client or target call.

Mapped capabilities

4 capabilities

Report assembly and structure
Producing a full diligence report organized by workstream from the ingested data room.
Risk flagging and severity
Surfacing issues such as non-competes, liens, or restrictive covenants and distinguishing material from routine findings.
Buy-side vs. sell-side framing
Adjusting emphasis and recommendations depending on which side of the transaction the user represents.
Company history summarization
Reconstructing an entity's corporate history from formation documents, consents, and charter amendments.

Deal Artifact Generation

Generating the recurring work product of a transaction: what to ask for, what to disclose, and how to respond to the other side.

“Full diligence reports, automated request lists, disclosure schedules, and more — in minutes, not weeks.” www.marveri.com

Mapped capabilities

3 capabilities

Automated diligence request lists
Producing a request list scoped to transaction type and to what is already present in the data room.
Disclosure schedule drafting
Populating schedules from underlying documents against the representations they support.
Responding to buyer or investor requests
Mapping an incoming request list onto available documents and identifying what still must be produced.

Completeness Audits & Reconciliation

Checking a corporate record for gaps and internal inconsistencies rather than reading any single document in isolation.

Mapped capabilities

4 capabilities

Cap table tie-outs
Reconciling a stated cap table against underlying stock issuances, option grants, and board consents.
Missing document audits
Identifying absent documents implied by others, such as a referenced amendment or an unexecuted assignment.
Document comparison
Diffing versions or related agreements and reporting the substantive changes between them.
Corporate housekeeping checks
Detecting stale good-standing certificates, unsigned consents, and unauthorized share counts in ongoing maintenance.

Illustrative example

Input: Tie out the cap table against the corporate records in this data room and list any equity issuances that lack supporting board approval.
Expected behavior: Reconciles each cap table line to an underlying issuance or grant document, and reports the option grant that has no corresponding board consent as an open item. Does not treat unmatched rows as approved or silently omit them.

Verifiability, Confidentiality & Trust

The guarantees Marveri sells alongside the output: every result traceable to a source document, and confidential deal data handled under stated security commitments.

“Marveri gives you a full picture on day one of data room access.” www.marveri.com

Mapped capabilities

4 capabilities

Source citation and traceability
Attaching a specific file and location to each finding so a reviewer can verify it.
Abstention on absent evidence
Declining to assert a fact the data room does not support instead of inferring or fabricating it.
Confidentiality and data handling posture
Accurate answers about SOC 2, GDPR, encryption in transit and at rest, and no training on customer data.
Cross-matter data separation
Keeping one deal's documents and findings out of answers about another matter.

Coverage is mapped from Marveri'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 Marveri test?+

The coverage map is generated from Marveri's own public product surface (AI due diligence platform for M&A and venture financings (legal tech)): 6 scoring areas — Data Room Ingestion & Organization, Contract Review & Clause Analysis, and Diligence Report & Issue Flagging, and more — spanning 23 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Marveri evals scored?+

Every case generated for Marveri — across Data Room Ingestion & Organization and Contract Review & Clause 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 Marveri library include?+

The full Marveri library is built on request. The coverage map spans 6 areas and 23 capabilities (for example, Document type classification and Naming convention normalization under Data Room Ingestion & Organization); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

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

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