Coming soon

Coming soon · AI due diligence for private equity

Every number, interrogated.

We are building the tool that turns a chaotic data room into structured, investor-grade output — running the actual math, cutting the cohorts, and flagging what the seller buried. Early access opens to a small group of funds first.

project_halyard / QoE / revenue_quality.xlsx Product preview

Net revenue retention by cohort

'19
'20
'21
'22
'23
'24
Source → DataRoom/03_Financials/Cust_Detail_2019-2024.xlsx, tab Billings, rows 4–2,118. Recomputed from invoice-level detail; management's NRR bridge excluded 3 churned logos.

Flagged in this run

Adj. EBITDA bridge ties$41.2MVerified
Top-10 customer concentration58.4%Flag
Gross margin, LTM vs FY22−410 bpsFlag
One-time add-backs$3.9MReview
Deferred revenue rollforwardTiesVerified
Headcount vs payroll register±0.0%Verified
Middle-market buyoutGrowth equityQuality of earningsCredit fundsSearch fundsCorp dev

What we are building

Pass sooner.
Commit with conviction.

Most tools summarise a CIM and call it intelligence. Sundooq is being built to do the work an associate would do at 2am — and to show every step of it.

01 / Speed

Kill bad deals in an afternoon

Surface concentration risk, margin compression and cohort decay on day one of due diligence — before you've burned a single advisor hour.

02 / Leverage

Twice the deal flow, same team

Data cuts, normalization, anomaly detection, schedule building and formatting — automated end to end, in your firm's template.

03 / Rigor

Real math, not a language model guessing

Every figure is computed in a deterministic engine against source rows. No hallucinated totals, no confidently wrong bridges.

04 / Defensibility

Every output traces to a cell

Click any number to land on the file, tab and row it came from. Excel-ready exports and a complete audit trail for IC.

Security posture

Your data room
stays yours.

Sundooq was built for institutions that cannot afford a maybe. Files are processed in an isolated tenant, never retained after a deal closes, and never used to train anything.

Request the security brief
No training on your data
Zero data retention
End-to-end encryption
SOC 2 readiness underway
Single-tenant deployment
Full access logging

Built by investors

We built the tool we kept wishing existed at 3am.

Sundooq started in a deal room, not a lab. The name is the Arabic word for the vault a fund is built on — and our founding team spent a combined nineteen years in middle-market buyout — rebuilding management's revenue bridge by hand, chasing a $600K variance across four versions of the same workbook, learning the hard way which add-backs never survive contact with the buyer.

That work is valuable. Doing it manually is not. So we encoded it: the cuts a good associate makes without being asked, the checks a good partner runs before IC, the traceability a good lender demands afterward. What comes out the other side isn't a summary — it's a schedule you can defend line by line.

The result is a platform that mirrors how your firm already works rather than asking you to work like software. No copilot theater. No confident nonsense. Just financial due diligence, done properly, at a speed that wins deals.

Talk to the founders

How it will work

Three steps from
data room to schedule.

No integration project, no IT ticket. Send files the way you already receive them.

01 / Ingest

Drop in the data room

Excel, PDF, scanned contracts, a zip from the banker's portal. We parse to row level and reconcile against the trial balance.

02 / Compute

Real math runs

Cohorts, retention, margin bridges, add-back verification and working capital — recomputed in a deterministic engine, not estimated by a model.

03 / Defend

Output you can take to IC

Flagged findings ranked by EV impact, an Excel-ready schedule, and every figure clickable back to its source cell.

Before you ask

The questions
every fund asks.

How is this different from an LLM summarising a CIM?

Summarisation is not due diligence. Sundooq recomputes figures from source rows in a deterministic engine, so a total is arithmetic rather than a prediction. Where a language model helps — reading contract clauses, classifying line items — its output is verified against the underlying data before it reaches you.

What happens to our data?

Files are processed in an isolated tenant and are not used to train any model. Retention is configurable per engagement, including delete-on-close. Ask us for the current security brief and subprocessor list before you send anything.

How long will a first run take?

Our target is a typical mid-market data room parsed in under an hour, with a full revenue-quality cut and flagged findings back the same day. Design partners will hold us to that.

Does it replace our QoE provider?

No. It removes the manual rebuild work your team and your advisors do before the judgement calls start, and it gives you a defensible position going into those conversations.

When can we actually use it?

We are opening access in stages to a small group of design-partner funds. Joining the waitlist puts you in that queue — we will come back to you with a date and, if you want, run a recently closed data room so you can judge the output against a deal you already know.

Early access · Opening in stages

We are opening to a small group of funds first.

Sundooq is not publicly available yet. Join the waitlist and we will come back to you with a date — and, if you want it, run a recently closed data room so you can judge the output against a deal you already know.

No newsletter, no drip sequence — one email when your access is ready.