'/%3e%3crect%20width='440'%20height='260'%20fill='url(%23gl4)'/%3e%3cg%20fill='none'%20stroke='%237e92cf'%20stroke-width='2'%20opacity='0.4'%3e%3ccircle%20cx='150'%20cy='96'%20r='10'/%3e%3ccircle%20cx='150'%20cy='132'%20r='10'/%3e%3ccircle%20cx='150'%20cy='168'%20r='10'/%3e%3c/g%3e%3cg%20fill='%239fb2e8'%20opacity='0.22'%3e%3crect%20x='176'%20y='91'%20width='104'%20height='7'%20rx='3.5'/%3e%3crect%20x='176'%20y='127'%20width='86'%20height='7'%20rx='3.5'/%3e%3crect%20x='176'%20y='163'%20width='96'%20height='7'%20rx='3.5'/%3e%3c/g%3e%3cpath%20d='M145%2096%20l3.5%203.5%206.5-7.5'%20fill='none'%20stroke='%235fe0d6'%20stroke-width='2.2'%20stroke-linecap='round'%20stroke-linejoin='round'%20opacity='0'%3e%3canimate%20attributeName='opacity'%20values='0;1;1;0'%20keyTimes='0;0.12;0.92;1'%20dur='5.4s'%20begin='0s'%20repeatCount='indefinite'/%3e%3c/path%3e%3cpath%20d='M145%20132%20l3.5%203.5%206.5-7.5'%20fill='none'%20stroke='%235fe0d6'%20stroke-width='2.2'%20stroke-linecap='round'%20stroke-linejoin='round'%20opacity='0'%3e%3canimate%20attributeName='opacity'%20values='0;1;1;0'%20keyTimes='0;0.12;0.92;1'%20dur='5.4s'%20begin='0.7s'%20repeatCount='indefinite'/%3e%3c/path%3e%3cpath%20d='M145%20168%20l3.5%203.5%206.5-7.5'%20fill='none'%20stroke='%235fe0d6'%20stroke-width='2.2'%20stroke-linecap='round'%20stroke-linejoin='round'%20opacity='0'%3e%3canimate%20attributeName='opacity'%20values='0;1;1;0'%20keyTimes='0;0.12;0.92;1'%20dur='5.4s'%20begin='1.4s'%20repeatCount='indefinite'/%3e%3c/path%3e%3c/svg%3e)
Most legal-AI tools can summarise a contract. The harder questions for M&A due diligence are whether you can trust and defend the output, and whether it fits your jurisdiction. Here's a buyer's checklist.
Judge the output, not the demo
Most tools look impressive in a demo. The question for M&A due diligence is narrower: can you trust the result, and can you defend it? Start with the deliverable. Do you get a chat answer to copy out, or a structured red-flag report, organised, prioritised and ready to act on? The shape of the output tells you what the tool was built for.
Can you trace and verify every finding?
For professional use, auditability is non-negotiable. Every finding should link back to the exact source document and paragraph in one click, so you, or a client, or an investment committee, can verify it rather than take it on faith. If you can't see why the tool said something, you can't stand behind it.
How does it handle accuracy?
Ask how the tool manages the risk of confident-but-wrong output. Good answers involve cross-checking findings, applying confidence thresholds, and flagging uncertain items for human review rather than presenting everything as settled. The honest framing is hallucination-resistant by design (built-in checks), not a promise of perfection.
Is it built for your jurisdiction?
Reading many languages is not the same as understanding a jurisdiction's law. A clause routine in one country can be a red flag in another, so ask whether the legal method is grounded in the actual law where your deal sits, native to local law, and authored by experienced practitioners, not generic prompts.
Where does your data live, and is it trained on?
Confidential deal data deserves clear answers: where it is stored, where it is processed, which regimes apply (for example nFADP and/or GDPR), whether there's a DPA, and whether your data is ever used to train models. Vague answers are themselves an answer.
Does it fit how your team works?
Finally, fit. Is the workflow built around an output you can act on, or a chatbot you have to drive? How does it handle open questions to the other side, and re-running the analysis as answers come in? A tool that matches how deals actually run saves more time than raw speed alone.
- ✓ Judge the output you get, not the demo.
- ✓ Insist on one-click traceability to the source.
- ✓ Ask how it manages accuracy: cross-checks, confidence thresholds, human review.
- ✓ Check jurisdiction depth and data protection (storage, processing, DPA, no training).