North Arm turns a deal’s data room into something you can ask questions of — from Claude, ChatGPT, Excel, PowerPoint, or the North Arm app. Every answer comes back with the file, the page, and the cell it came from.
A retrieval engine built for M&A. Point it at a data room export and it reads every file: PDFs and scans, spreadsheets down to the cell, contracts down to the clause, emails and their attachments. It builds an index that Claude and ChatGPT can search, and keeps that index isolated per deal.
It is not a data room. It does not replace Datasite or Intralinks, and it does not hold your documents. It is not another chat window to check. It plugs into the assistant your firm already pays for.
Drop the export ZIP into North Arm or point it at the folder. North Arm reads each file once to build the index and discards it in seconds. It never keeps a copy.
Minutes later you have a data room index in Excel: every file, a one-line description, a detailed description, and a link that opens the file on your own drive.
Add North Arm to Claude or ChatGPT once. Then ask in plain language — in the app, in Claude, in ChatGPT, or inside Excel and PowerPoint. Answers carry citations. If it is not in the room, North Arm says so instead of guessing.
“What is the NWC peg in the latest SPA draft, and does it tie to the Q2 balance sheet?” You get the answer, the file, and the page. Not a list of documents to open.
Pull figures into the model from Claude or ChatGPT with the cell they came from. Or ask for a preloaded workbook: staged values in one tab, and a Sources tab with a screenshot of each PDF source beside the number it feeds.
Drop the new export. North Arm shows what is new, changed, or gone, and answers “what’s new since yesterday’s call.”
Upload a few of your own decks once. North Arm learns the layout and builds teasers, CIM summaries, and buyer updates on your master, with a citation footer on every slide and a sources slide at the end.
Hand it the teaser, the CIM, and the room. It proposes a plan, drafts an outline, waits for your approval, then builds the deck and preloads the model so you start with the inputs in place.
Every room gets its own address. Forward a file to it and the file is indexed and answerable minutes later.
A question list from Excel becomes a matrix of documents by questions, every cell cited. Memos draft from your templates, section by section.
Legal hold when you need it. When the deal closes, one action destroys the index and produces a closing package from the access log, written to be read by a risk team.
North Arm keeps an index you can delete, never the documents.
Claude runs on your firm’s own Anthropic account, on your terms and your retention settings. Your data is never used for training and never sent to a model you did not choose.
Who asked, from which surface, which files came back, and when. Never the text of the question or the answer. Conversations in the North Arm app are saved with the room and go when the room goes.
Nothing crosses between deals.
Sign in with your firm’s SSO. North Arm reads the room. It never writes to it.
Their conversations reach the room through North Arm; the documents are never in their file storage. Delete the room and there is nothing left to find.
Repeat founders. One spent five years in M&A and closed $2B of transactions; the first version of North Arm was built on live deals because the data room was the slowest part of every one. Now they’re scaling it.
Every deal team is running AI on the room now. Theirs has read every file, sourced every model input, and sharpened every assumption before the first call. They are faster. Their models are more accurate. They understand the business, and its risks, better than you do — because the data room stopped being a mystery to them.
You’re booked.
We’ll bring a sample data room — bring yours if you can. A confirmation is on its way from a person.