Today Centari is releasing Agentic Review in preview, which shortens the path to verified deal data, together with updates that put that data to work in live negotiations and from your email inbox.
In a demo, AI platforms for deal work look alike: a chat window, a grid, fields pulled from a contract. Every firm and fund now has capable AI tools in its tech stack, and for ad hoc work they are useful. The real test comes after the demo. At a law firm, a partner has to stand behind the output in front of a client. At a fund, the output has to hold up in an audit.
AI outputs that appear in seconds still have to be checked. When no one checks them at the source, every end user has to check them again, on every query and for every downstream use case. We built Centari to check once, at the source, so the data becomes an institutional asset that can be relied on everywhere. Unverified data is a liability. Verified data is an asset. Agentic Review gets your team to the asset sooner.
A data asset you can trust starts with how Centari reads a document. Deal documents are systems of logic. A defined term on page 12 controls a covenant on page 47. A side letter changes what a partnership agreement means for an LP. An amendment two years later modifies key provisions that came before it. Centari's Deal Reasoning Engine maps how a document is built before extracting anything, and reasons across every related document in the deal, amendments and side letters included. Every data point Centari returns is cited to its source language.
Our frontier document understanding capability is why the most demanding transactional practices in the world, including Ropes & Gray, Willkie Farr & Gallagher, Fried Frank and Wilson Sonsini, run their deal intelligence on Centari. And the results are measurable. At leading Australian law firm Gilbert + Tobin, analyzing a deal with 200 data points used to take a full day and now takes under an hour.
Trusted data, faster
We built Centari around human-in-the-loop verification, combining industry-leading data quality with a streamlined interface to audit AI outputs. Agentic Review, now in preview, keeps the human in the driver’s seat while removing time-consuming repeated checks. Centari recognizes patterns in how your team approves outputs and can now auto-verify high-confidence answers, freeing reviewer time to focus on just the answers that require a judgment call.

Administrators have complete control over where Agentic Review runs, and every automatic verification stays in the audit log. Every review your team completes makes the next one faster, and every verified deal builds AI-ready data your organization can rely on for years.
Your deal history, where decisions get made
Verified data pays off at the moment of a decision, and decisions rarely get made inside a database. Today’s release brings your organization’s deal history to the surface in two more places where decisions happen: live negotiations and your email inbox.

The markup from the other side. When a markup lands on your desk, the question is always the same: how far does this drift from what we've accepted before? Now you can upload the draft to Centari Intelligence and get a reliable answer based on your deal history. Lender’s counsel can go into the client call with specific examples of where a borrower's first turn of a credit agreement strays from terms accepted from comparable borrowers, with each point cited to a prior deal. A fund’s IR team can see how an incoming side-letter request compares with every side letter the fund has granted. Every answer draws on your own history.
The question in your inbox. For deal professionals, work happens in email. You can now access Centari Intelligence directly from your inbox without opening another system. The next time someone asks, "What indemnity caps have we accepted in our last ten mid-market SPAs?" simply forward the question to Centari and get an answer back for your review in seconds. Replies come only to you, under the same access controls and ethical walls as the platform.

Putting your institutional data to work across your AI stack
Your data should work for you across every tool your organization runs. Centari is built for interoperability. The Centari MCP Server connects your deal database to all applications that support MCP, the open standard for connecting AI tools to data, so an answer in a general purpose platform draws on the verified record in Centari.
Every organization can now buy capable AI. No one can buy your deal history. Private markets run on documents, and we're building the information infrastructure to put every one of them to work.
To see what your own documents can do, request a demo.




