Crossfire

Northwind Labs, FY2025 · pre-audit substantive testing

Two agents test the books and a person rules only on what they cannot settle between them. Every answer cites a row in the books, and every ruling is remembered, so the next run starts where this one ended.

Auditor
Samples the books risk-weighted and demands evidence for each pick: three-way match, cutoff, bank rec, revenue tie-out.
Accountant
Searches invoices, contracts, the bank feed, Dodo Payments, and the ledger. Cites rows or admits the gap.
Controller
A person. Rules on what escalates: sufficient, needs more, accepted with note, or exception with a remedy. Rulings become memory.
Samples
25
Defended
23
Gaps waiting
0
Exceptions
2
By memory
3
Coverage
92%

23 of 25 defended, 3 of them by memory · +4 pts since run 1

Since the last run

Same books. Every ruling the controller filed on run 1 was on the accountant's desk for run 2.

Coverage
88%92%+4 pts
Gaps left
32
Resolved by memory
03
Samples
2525

Audit runs

Newest first. Seed, materiality, sample size, and cycles fix the sample, so the same inputs draw the same items every time.

RunCreatedSamplesDefendedGapsExceptionsOpenCoverage
2FY2025 substantive testing, pass 22026-09-06 21:162523020
92%
1FY2025 substantive testing, pass 12026-09-06 21:132522210
88%

The books

Seeded deterministically. Ten planted issues, everything else reconciles.

Vendors
5
Contracts
5
Invoices
61
Bank transactions
300
Dodo transactions
437
Ledger entries
1,586