The AI-native data room of recordAmsterdam · New York · London
Articles
Articles from the CogniSuite team on M&A market data, the rules that apply to deal software, and how our own product handles documents, permissions and AI. The market pieces stay inside what published sources support and say plainly where the trackers disagree with each other. The product pieces describe what the software does, evidenced from the code, and where each capability stops.
What a FIPS validation certificate covers and what it leaves out, why the 22 September 2026 move to the Historical List is a reason to reopen the question, and seven questions that separate a real answer from a phrase.
Which of the two Article 50 transparency duties falls on the platform and which falls on your deal team before 2 August 2026, and what to ask a vendor.
The quarter set records by value while transaction counts fell to levels not seen since 2010. The data is still preliminary and the trackers do not agree with each other, which is itself the useful lesson for anyone running a diligence process.
Six cybersecurity agencies published joint guidance on 1 May 2026 advising organisations not to grant AI agents broad or unrestricted access to sensitive data. A due diligence data room is one of the most concentrated collections of sensitive data a firm will ever assemble, and the access inside it is deliberately unequal. That makes the connection between an assistant and a room a permissions problem before it is an AI problem.
What happens on the request side when several bidders ask the same question in the same week, from the drafted answer through to the person who approves it.
How much confidential material one compromised deal room account can reach, and what per-deal separation and per-folder access grades do to that number.
What amended Regulation S-P requires a covered firm to be able to evidence about a data room vendor under the 3 December 2025 and 3 June 2026 compliance dates.
What the Q1 2026 data agrees on, where the trackers contradict each other on deal counts, and what a handful of very large transactions meant for diligence work.
What the run of US court sanctions over AI fabricated citations implies for AI drafted diligence answers, and what verifying a quote against its source still does not catch.
What searching the room by meaning finds that a filename search misses, and where the method is weaker on scanned pages, spreadsheets and very long files.
Why Q4 2025 closed the strongest year since 2021 on fewer and larger deals, and why the published full-year totals differ by hundreds of billions of dollars.
How an arbitrary Excel diligence checklist becomes a working request list when the model is asked for a reading plan and ordinary code copies your cell text.
How the AI is restricted to documents the person asking can already open, why that filter runs before ranking rather than after, and what it does not cover.
How a diligence request gets a written answer with verified quotes attached, and how an answer already given elsewhere in the deal can be carried across instead of written again.
What the platform checks about a money figure in an AI drafted answer, and why that is not the same thing as reconciling numbers across every document in the room.
How one room serves the counterparty, your own side and a clean team scoped to one named organisation, with the same permission rules governing the AI.
How every quote in an AI drafted answer is checked against the stored document before anything renders, and what happens to an answer whose evidence fails.