Preclose: tax due diligence, from request list to report
Preclose is the platform we are building to take the manual work out of tax due diligence. Rules do the calculating, people do the judging, and AI explains the results in plain language.
How a deal moves through Preclose
-
Request
Send the target a structured information request list. Each item is tracked as outstanding, received or queried, with no email threads to reconcile.
-
Extract
Uploaded returns, assessments, trial balances and agreements are read and the tax data is pulled into a consistent model.
-
Compute
Jurisdiction rule packs calculate exposures, penalties and interest. The same inputs always give the same answer, and every figure links back to its source document.
-
Report
A draft report is assembled with findings, quantified exposures and suggested SPA protections, ready for the adviser to review and sign off.
Why the maths is not left to AI
A due diligence number ends up in a purchase agreement. It has to be reproducible and explainable to a tax authority years later.
So Preclose computes exposures with deterministic rule packs for each jurisdiction, versioned like code. Language models are used only to summarise findings and answer questions about them, always citing the computation they describe.
Who uses it
- Acquirers and deal teams
- See exposures as they are quantified and feed them into price, indemnities and escrow discussions.
- Advisers
- Spend time on judgement rather than chasing documents, and keep a full audit trail behind every conclusion.
- Targets
- Answer one clear request list through a secure portal instead of repeated spreadsheets.
Initial jurisdiction coverage: Middle East, Africa and the United Kingdom.