Marveri

What it is

Marveri is a diligence platform for transactional lawyers that takes an entire data room as input and returns the work product a deal team owes: an organized file set, a diligence report, a request list, a disclosure schedule, and a cap table tie-out. Founded in 2023 and run from Cambridge, Massachusetts, it was built by Connor Acle — a former Morrison & Foerster corporate associate — with MIT AI researchers Emily Mu and Samuel Bosch. It came out of stealth in May 2025 on a $3.5M seed round reported by Law360 Pulse.

The distinguishing product decision is that there is no chatbot. You do not prompt Marveri and you do not ask it questions. You point it at a target’s data room and it renames and folders every file to your own naming convention, runs a group comparison across the whole set to find which agreements are off-template, flags what is unexecuted or missing, and drafts the report with every assertion linked back to its source document. The vendor’s framing — no prompting, just cited results — is a positioning line, but it is also an accurate description of the interface.

Why it shows up in Legal Ops stacks

• It is scoped to deals, not to legal work. Kira, Luminance and Harvey all sell across practice areas. Marveri covers M&A, venture financings and corporate records, and the company has said it is not expanding past them. For a firm whose diligence pain is real and whose research pain is not, that narrowness is the reason to buy rather than a limitation.
• The unit of work is the data room, not the document. Comparison runs across the full set at once — 2 documents or 20 — so “which of these 140 NDAs deviate from our form” is one operation instead of a review queue. Off-template detection is the load-bearing feature here; clause extraction alone is table stakes by 2026.
• Both sides of the deal, one tool. Sell-side, it organizes the client’s own records, finds the missing signature pages before the buyer does, and answers the request list. Buy-side, it ingests the room and reports. Most diligence AI is built for the reviewing side only.
• Setup is a folder connection, not a rollout. Google Drive, SharePoint/OneDrive and local folder sync are the ingestion paths; Word and Excel export carry the citations back out. No DMS migration and no professional-services engagement — the specific cost that keeps small deal shops off Litera.
• Cited output is the audit artifact. Every line in a Marveri report resolves to the underlying page, which is the standard the partner signing that report needs. See grounding vs hallucination in legal AI.

Pricing reality

No published price, no pricing page, no self-serve tier — quote only, gated behind a demo. Checked 9 September 2026. That is a cost of evaluation, not a formatting complaint: you cannot size this against a budget without a sales conversation, and a two-partner shop ends up negotiating with no reference point.

The comparison set is the anchor. Third-party survey data (ILTA 2025) puts Kira in the $25,000–$85,000 per year band for firm deployments, and our own Luminance entry puts mid-market deals in the low five figures with enterprise rollouts reaching six and seven. A seed-stage vendor selling to boutiques that were never Kira prospects has to land under that band to win, and the customers named on its own site — Optimal Counsel, Darwin Legal, Nexus Venture Counsel, Duncan Bergman Mandell, Proviso — are small firms rather than AmLaw names. Ask for per-matter or per-data-room pricing before you accept a per-seat annual; the workload is deal-shaped, and the meter should be too.

Security is SOC 2 Type II with encryption in transit and at rest, and the vendor states customer documents are not used for training. Get the no-training term into the order form — a website line is not a contract.

Best for

Boutique M&A firms, startup and venture counsel, and small corporate-development teams that run real deal volume, need the diligence report and the disclosure schedule moving on day one of data room access, and cannot justify a Litera platform commitment or a 60-90 day implementation. It is also the sell-side prep tool for a company organizing its own corporate records ahead of a raise or an exit.

Alternatives and when to pick them

• Kira Systems — the diligence incumbent, now inside Litera. Pick it when you already run Litera Transact and Compare, when the bundle economics beat a standalone, or when adoption across 40+ practice areas matters more than the deal workflow itself. Its deployment across most of the AmLaw 100 is also the answer to “what will the other side be using.”
• Luminance — pick it when the work extends past diligence into negotiation and review as an ongoing function rather than a deal event. Its pilot-first motion on a single contract type is the better evaluation path if you are still unsure the category works for you.
• Harvey — at an $11B valuation as of March 2026, the fastest-growing entrant and the one with the most enterprise distribution. Pick it when the firm wants one AI surface across litigation, advisory and transactional work, and will accept diligence that is good rather than specialized.
• Legora — $5.6B post-money in April 2026, 1,000+ firm and enterprise customers, $100M+ ARR. Pick it for the same firm-wide scope as Harvey when European coverage or the collaborative drafting surface decides it.
• Ontra — pick it when the recurring pain is high-volume routine contracts (NDAs, side letters) inside a fund rather than episodic deal diligence.

If your deal volume runs under about one transaction a quarter, none of these clear the bar. Contract associates and a shared drive are cheaper.

Watch-outs

• Seed-stage vendor in a deal-critical position. $3.5M raised against Litera, Luminance and Harvey. Guard: negotiate an export clause at signature covering both the organized file structure and the citation-linked reports. You already own the documents — make sure you own the structure.
• No published accuracy benchmark. The vendor claims cited, verifiable output but publishes no measured error rate. Every result being clickable is not the same as every result being right. Guard: run it against a closed deal where you already know the findings, and count what it missed rather than what it found. The method is in AI contract review accuracy.
• “Days, not weeks” is a vendor claim. The one quantified case study on the site is a 70% reduction in client onboarding time, which is a different job than diligence turnaround. Guard: measure associate hours on your first two matters against the last comparable deal you ran by hand, and do it before the renewal date.
• No DMS, no API, no MCP. Ingestion is Drive, SharePoint/OneDrive and folder sync. If your records live in iManage or NetDocuments, every matter carries a copy step in and out. Guard: price that copy step in associate time, and get the DMS roadmap in writing before committing past one year.
• Quote-only pricing with no anchor. Guard: get two competitive quotes — one for Kira through Litera, one for Luminance — before your first Marveri call, so the number you hear has something to sit against.

For the surrounding category, see Kira Systems vs Luminance and best AI tools for legal ops.