Reading a data room requires practitioners who understand commercial risk, deal structure, and the specific vulnerabilities that matter to their client. However, and notably, before any of that judgment can operate, someone has to read several hundred contracts. In a competitive process, the (unlucky) person tasked with that job usually has less time than the task requires.
The result of this phenomenon is familiar. Contract reviewers prioritise by order of importance, and material provisions hidden in lower-priority contracts get missed. These material provisions can include change-of-control clauses, IP ownership claims, or assignment restrictions. The issue here is less the practitioners’ ability to look for these material provisions, but more so the fact that the data room is so extensive that they cannot get through everything in a timely manner whilst still using an eagle eye for spotting these clauses.
Post-deal surprises are among the most expensive problems in transactional practice. For instance, a change-of-control clause that triggers on acquisition without the counterparty's consent can have significant consequences like the termination of a key commercial agreement. These consequences don't surface until after completion of the deal, at which point the buyer has limited leverage and full exposure. The reason these consequences weren’t flagged earlier is often because manual review under time pressure is, by nature, not an exhaustive process.
Sell-side disclosure schedule preparation is the other side of the same problem. The sell-side team has to read across the target's contracts, financial records, litigation history, regulatory correspondence, and property documents, identifying everything that qualifies as an exception to the warranties being given. Missing a disclosure creates warranty liability after completion, while including too much creates negotiation problems during the deal. The task requires precision across a large body of materials, and the time available for it is set by the deal timeline, not by the volume of material.
The issue from the examples offered becomes clear: the cause of post-deal surprises is often a lack of capacity to get through materials with precision at scale, rather than being a problem with the practitioner’s professional judgment.
AI has made a specific dent here. Not in due diligence broadly, but in these two tasks. Both require reading across a large corpus and surfacing specific provisions, with no deal context required and no judgment call involved. They're speed-limited research tasks, and that's where AI has the clearest advantage.
A change-of-control review that used to take three practitioners two days can be turned around in hours, with the AI's output reviewed against primary documents by a single practitioner. The time saving is real and consistent across transaction types and deal sizes. Unlike most efficiency claims made about AI in legal practice, this one is measurable at the matter level.
Major transactional firms globally are now formalising what practitioners have been discovering. Sullivan & Cromwell published guidance on AI use in M&A transactions in January 2026. Thomson Reuters acquired Noetica, a startup building AI tools specifically for corporate transaction analysis, in February 2026.
Very recently,Morgan Lewis framed the industry shift in March 2026 as moving from competitive advantage to governance imperative: referring to AI as a ‘core pillar of corporate strategy’.
Legal AI in the M&A use case specific thus offers a clear return on investment, that is concentrated in high-volume tasks where speed is the binding constraint and precision is the measure of quality. In instances like mitigating against post-deal surprises, and sell-side disclosure preparation, Legal AI has a clear benefit for M&A practice.
In an auction process, the buyer who gets through the data room faster can price risk more accurately and bid with more confidence. In a competitive process where multiple bidders are reviewing the same documents, that's a real structural advantage. A few days of additional certainty about the risk profile can change a bid. The ability to tell a client, mid-process, that you've already cleared the data room and compiled a report of the results is a different kind of quality of service than still being in the weeds of the documents when the vendor's deadline arrives.
When document review compresses, practitioners also recover time for the work that actually requires professional judgment: the risk assessment, the analysis of what was found, the client conversation about how to price or respond to a specific issue. AI shortens the path to that point.
An AI system flagging change-of-control clauses across a data room still needs a practitioner to review the flagged provisions against the primary documents. Liability for missing a material provision doesn't transfer to the software. That's always been true of junior review too. What changes is how quickly the first pass happens, and how complete it is when it reaches the practitioner making the call.
The practitioner who used to spend three days reading sequentially now reviews a curated set of flagged issues before making the judgment calls the matter actually requires.
The firms extracting the most consistent value from AI in transactional practice designed this workflow deliberately. They mapped where AI sits, what it produces, and what the review step looks like. The technology is available to almost everyone: what differs is whether people design it for where it fits.
Habeas grounds every output in verified Australian legal sources with clickable citations, so you can get to verification faster. Book a demo.
