TL;DR
- Contract review is roughly 30% of the time spent on a diligence. The other 70% is coordination: allocating work, tracking a moving data room, running Q&A, and building the report.
- Ask any associate what kills a deal, and it is never the reading. It is checking junior work, working out what has changed, and assembling the report.
- If your AI tool only speeds up the 30%, you have bought a faster engine for a car with no steering wheel.
The bottleneck nobody is fixing
Every AI vendor sounds the same. Contracts processed faster. Issues flagged sooner. Juniors freed from the first pass through the data room. All true, all valuable, and all beside the point.
If AI can scan a data room in hours, where do the next six weeks go? We've all had general purpose tools for months to years now. Clients are rightly asking why their bills are staying the same, and why diligence still takes as long to deliver.
The time doesn't go to reading. It is burned during orchestration: deciding who looks at what, chasing what is done, reconciling what comes back, and turning it into one document a partner will put their name on. None of that is a legal problem, which is exactly why most legal AI doesn't touch it. If doc review is a slice, coordination of the deal is the whole pie.
What the other 70% actually looks like
Specialists are allocated documents by spreadsheet. By Wednesday that spreadsheet is old news. One specialist finished early and quietly picked up documents nobody formally assigned. Two lawyers have read the same contract and billed for it, because it sits in the data room twice under different names.
Overnight, the data room updates. Most platforms will tell you what is new. Almost none tell you what was modified or quietly removed, so nobody knows which reviewed documents changed underneath them. We know on deals that running underneath all of it is a Q&A log with hundreds of live questions.
- The same point was raised by three different specialists.
- A client capping how many questions you can send, so someone has to triage which ones are worth the quota.
- Answers that do not answer the question.
- A scramble to route responses back to whoever asked before the next draft goes out.
- Then a partner asks where the deal stands, and a senior associate burns half a day finding out.
The endgame is where the write-offs live
Then comes assembly. One senior associate pulls everyone's work into a single report: reconciling formats, judging how far to soften each specialist's findings, rebuilding it in the template the firm actually sends, chasing defined terms and cross-references until it reads like one person wrote it.
Our customers tell us this takes 20 to 40 hours, and it happens more than twice on a deal. It routinely takes longer than the document review it is summarising. It is invisible, hard to bill, and it is where the write-offs come from.
What the best deal teams do differently
They are not trying to read faster. They are pairing purpose-built diligence AI with real uplift on deal management and coordination.
- One live view of the deal. Allocations, outstanding items, and progress visible to everyone, so nobody has to send a status email to find out where things stand.
- Change tracking that works. When the data room moves, the team knows exactly what moved. No blanket re-reviews, no VLOOKUPs.
- Report assembly built in. Templated to the firm's house style and exported to Word in one click.
Diligence is an exercise in precision under time pressure. That makes it a workflow problem, and workflow problems are not solved by bolting a chat window onto a document reader.
The best questions to ask in a demo
Do not just ask how fast it reads. Ask what happens to the other 70%.
Who allocates the work? Who catches the data room changes? Who builds the report? If the answer is still a spreadsheet and some poor soul, you have not rethought diligence. You have made one slice of it incrementally better, and in our books that is not enough.
Deeligence was built by M&A lawyers who have run these deals themselves. Our AI is pre-trained, so it works out of the box and does not cost you a week of setup on a live matter. Around it sits an orchestration layer that keeps the whole deal team moving in unison.
The point was never to read the data room faster. It was to get the deal done.