Deterministic engine
No AI in payroll math. Runs are priced by code and rule snapshots, with tests guarding invariants.
Runbook has to earn trust twice: by proving the payroll record and by protecting the people, money, tax data, and approvals inside it.
Wrong checks, late raises, tax notices, mid-year switches, garnishments, and worker questions all collapse into the same practical problem: what happened, why did it happen, and what evidence proves it? Large payroll companies prove trust with scale, security programs, guarantees, and support teams. Runbook needs those operational controls too. The product adds something more direct: a payroll run that can explain itself, a proof packet another party can verify, and a live-funds gate that blocks money movement until the security and operating posture is ready.
Buyers need familiar security signals and a deeper payroll-specific proof model. Runbook should show both plainly.
These are already part of the product architecture or repo, not marketing hopes.
No AI in payroll math. Runs are priced by code and rule snapshots, with tests guarding invariants.
Money is represented as integer cents end to end, with display conversion only at boundaries.
Facts carry effective time and recorded time, so corrections can distinguish what was true from when it was learned.
Approved runs can be sealed and independently verified by recomputation, not trust in a dashboard.
Real rule values require ratified or explicitly deferred provenance; demo values must remain labeled.
Agents propose documents. Humans confirm. The engine validates. The ledger records.
Append-only entries, chained digests, public verification, and tamper-evident history without putting payroll data on-chain.
These are not marketing badges. They are product gates for payroll, filings, bank data, and worker records.
These public promises are intentionally ambitious. Each one is a build obligation before general availability.
The Quittance is the core trust object: source events, rule snapshot, calculation, accrued liabilities, settlements, canonical digest, signature, and registry status.
Verify a QuittanceRunbook's proof model is blockchain-like where it matters: append-only entries, chained digests, public verification, and tamper-evident history. But payroll data stays private. The blockchain is not the ledger; the ledger is Runbook's bitemporal payroll record. Public anchoring is a future optional witness, not the source of truth.