Compound runs due diligence the way a buy-side deal team does: it reads the whole data room, works each diligence stream into an Excel workbook with live formulas, and cites every number to the page or cell it came from. A deal team can drop in up to 15,000 files at once and start hundreds of tasks that subagents run in parallel. Each section below is one diligence task buyers bring to Compound, with a worked example run on a full data room.
Data room review against the request list
The first pass of AI due diligence maps the virtual data room against a standard buy-side request list. In the AI data room review of Project Voltaic, a fictional data center sale, Compound indexed 161 files against 68 standard requests: 23 are fully present and 17 partly, the financials stop at March 31, 2026, and the sell-side QoE is still a draft. Compound then drafted 23 requests for the banker, 7 of them urgent.
Quality of earnings: retesting the seller’s QoE
Compound retests a sell-side quality of earnings (QoE) report against the primary records: the audited accounts, the monthly management accounts, the billing export and the meter data. In the AI financial due diligence example, buy-side adjusted EBITDA is $206.9 million against the seller’s $224.0 million, and net debt and debt-like items are $77.1 million above the seller’s figure. At the 19.2x multiple the campus appraisals imply, those findings are worth about $405 million of equity value.
Adjusted EBITDA add-backs and the working capital peg
Compound tests each add-back in a seller’s EBITDA bridge against its evidence and records a verdict for every line. The adjusted EBITDA add-back test accepts, adjusts or rejects each of management’s adjustments and takes $224.0 million to $215.1 million of diligence-adjusted EBITDA. The net working capital peg example applies the QoE’s own definition of working capital, which its table does not follow, and moves the peg from $(11.8) million to $(16.4) million.
Fund due diligence: the LP’s DDQ review
For an LP, Compound tests each answer in a manager’s due diligence questionnaire (DDQ) against the rest of the fund’s data room. In the fund due diligence example, Compound rebuilt an illustrative $1.5 billion buyout manager’s track record from the cash flows: the 3.6x headline covers seven of 38 deals, the four funds return 2.32x gross, and Fund IV’s net IRR falls from 14.5% to 11.8% without its subscription line.
Real estate due diligence: the diligence log
On a property acquisition, Compound turns the data room into a diligence log with an owner, a question for the seller and a due date on every open item. In the commercial real estate due diligence example, the log for an illustrative industrial park holds 52 items and 19 questions for the seller, and on in-place income the findings support about $43.1 million against a $48.0 million ask.
Why buy-side teams run due diligence in Compound
- Every number traces to its source. Click a figure in a deck to open the model cells behind it, then a cell to open the data room page it came from; a downloaded workbook keeps the citations as cell notes.
- The whole data room at once. Drop up to 15,000 files at once into a deal’s agent, and every task in that agent can read them.
- Real Excel, PowerPoint and Word files. The diligence workbook downloads with live formulas, and the IC deck and memo come back in your firm’s templates.
- One agent for the deal team. An associate starts the work, a VP comments on a cell and the agent makes the change, and the partner opens the same file.
- Enterprise security. Compound is SOC 2 Type II, encrypts data with 256-bit AES at rest and in transit, never trains on customer data, and keeps each agent’s files apart, so one deal’s numbers never reach another deal’s analysis.
