APIR
The evidence record for AI agents
Numbers Check
6623641c-0d35-4f8a-8fb8-7b844dd80a49
Sealed, verified
Specimen. Fictitious figures, entered to demonstrate the record.
Numbers checked · 22 September 2026
Harbourlight Aged Care Services Pty Ltd (specimen, fictitious)
The figures declared for this business were checked for internal consistency by APIR and read as arithmetically inconsistent on the arithmetic tests of the Triage Verdict: 4 findings.
On 22 September 2026, Slav Ripa, for Harbourlight Aged Care Services Pty Ltd (specimen, fictitious), entered 3 years of revenue and net result as they appear in the lodged tax returns and the management accounts, and the year-end receivables and payer cycle, and fingerprinted 1 document in their own browser. The Triage Analyst ran the two arithmetic tests of the Triage Verdict on the figures as entered. The result is arithmetically inconsistent. Every finding and its workings are below.
The figures were declared, not audited. APIR did not read the documents and did not check them against any register. It recorded their fingerprints, so the copies you are handed can be matched to this record. This is arithmetic on whether the figures can all be true at once. It is not a valuation, not an audit and not advice, and it says nothing about whether the business is worth buying.
| The read | Arithmetically inconsistent The documents contradict one another and cannot all be true as presented. 2 of 2 tests run, 3 critical, 1 high. |
| Figures entered by | Slav Ripa |
| Tests run by | Triage Analyst, a papered agent under mandate MND-CHARTER-V1-P05: run the arithmetic checks, produce findings, produce no verdict |
| Findings approved by | Slav Ripa · 2026-09-22 10:36 UTC |
| Sealed and recorded | 2026-09-22 10:36 UTC · event 7 on the analyst’s chain |
| Outside clock | Chain head timestamped by an RFC 3161 authority at 2026-09-23 00:00 UTC |
Test 1 of 2, tax to book variance
Net result, share of management-account revenue| Year | Revenue, accounts | Revenue, return | Net, accounts | Net, return | Variance | % of revenue | Assessment |
|---|---|---|---|---|---|---|---|
| FY2023 | 1,980,000 | 1,952,000 | 58,000 | -141,000 | 199,000 | 10.1% | Critical |
| FY2024 | 2,140,000 | 2,118,000 | 96,000 | -37,000 | 133,000 | 6.2% | Critical |
| FY2025 | 2,310,000 | Not provided | 412,000 | Not provided | n/a | n/a | Critical |
Method. Net variance under 5.0 percent of revenue is acceptable. Between 5.0 and 10.0 is medium. Above 10.0 is high. A sign reversal between the return and the accounts in the same year is critical whatever its size. A missing return for a claimed year is critical. Revenue variance is shown for context and is not graded.
Test 2 of 2, cash cycle
Expected receivables against reported| Total billings, year | 2,310,000 | High |
| Billings on the slow settlement cycle (90% at 30 to 60 days) | 2,079,000 | |
| Expected receivables, low to high | 170,877 to 341,753 | |
| Receivables reported at year end | 118,000 | |
| Implied debtor days | 19 |
Reported receivables sit below the bottom of the settlement cycle. Either collections run faster than the cycle, or not all reported revenue was billed. Method: expected receivables are slow-cycle billings times settlement days over 365; implied debtor days are reported receivables over billings times 365. Outside the range is high.
The documents, by fingerprint
1 file, SHA-256, hashed in the declarant's browser| File | Size | SHA-256 |
|---|---|---|
| specimen-triage-verdict.pdf | 287.1 KB | ddbb15d99b6ededb732df4da1b0c32bbb292f5e79ecfbcf3c46cbaf25713c3dc |
A file that produces the same SHA-256 is the document the figures were declared from, byte for byte. APIR never held the files and did not read them.
Integrity, recomputed on this open
2026-10-13 10:19 UTC| Declaration and figures hash | 4f8eaa62eefffbcce2aa29a3ede7379bd355ec940d210b265caf5c67b7f8c652 | Recomputed, matches |
| Result hash | e6379043215051d35dacb06f342ce52b3fc2440407d45dea989178f180031e70 | Recomputed, matches |
| Engine rerun on the sealed figures | numbers-check-v1 | Reproduced the sealed result |
| Chain hash | 0f2ef2b7e0d3495bbd5ceb7684d3c2d7d5d7a6a2cef47c1a7e660d9285f338df | Chain intact, 28 events |
What this certifies
- That these figures, in these words, were entered by the named signed-in person at the time shown.
- That the two arithmetic tests were run on exactly those figures by the named agent, and produce this result whenever rerun.
- That a named person approved the findings before the seal, and the record has not changed since.
What this record does not confirm
- That the figures match the documents, or that the documents are genuine. APIR did not read them and did not check them against any government, bank or register record. The fingerprints let a reader match copies; they do not read them either.
- That the business is sound, worth its price, or worth buying. No valuation, no audit, no assurance, no advice, no opinion on performance.
- Anything about why a variance exists. Findings identify inconsistencies in figures, not the reasons for them.
- A passed test means the figures are consistent with each other. It does not mean they are true to the outside world.
Reading this record in thirty seconds
For the analyst it was forwarded to- What you are looking at. A sealed record of arithmetic tests on figures a named person declared, created 2026-09-22 10:36 UTC, tamper-evident since, timestamped by an outside authority.
- Do the declared figures agree with each other? The tests above answer it, with the workings. Arithmetically inconsistent, 2 of 2 tests run.
- Has this record changed since it was sealed? No. Both hashes recomputed and the tests reran to the same result on this open.
- Who is responsible for what? The figures were declared by Slav Ripa. The tests were run by Triage Analyst, a papered agent of APIR, and the findings were approved by Slav Ripa.
- The one question it does not answer. Whether the documents are genuine reflections of reality. No consistency check can answer that, and a service claiming to is overstating.
- What to do with it. Treat a finding as a question for the seller. Treat a passed record as one input to your ordinary diligence, never a substitute for it.
The mark, for a listing
Copy it as text or as a linkNumbers checked · 22 September 2026
The figures declared for this business were checked for internal consistency by APIR and read as arithmetically inconsistent on the arithmetic tests of the Triage Verdict: 4 findings.
Sealed and anchored by APIR. Anyone can open the record: https://apir.ai/numbers/6623641c-0d35-4f8a-8fb8-7b844dd80a49
As a link: <a href="https://apir.ai/numbers/6623641c-0d35-4f8a-8fb8-7b844dd80a49" rel="noopener">Numbers checked by APIR, 22 September 2026: The figures declared for this business were checked for internal consistency by APIR and read as arithmetically inconsistent on the arithmetic tests of the Triage Verdict: 4 findings. Sealed and anchored. Anyone can open the record.</a>
Scan or open to check
https://apir.ai/numbers/6623641c-0d35-4f8a-8fb8-7b844dd80a49
Recomputes both hashes, reruns the tests and checks the chain on every open, no account needed. It proves the record is intact. It does not judge the business.
APIR, operated by RIPA PTY LTD, ABN 48 638 569 632Checkapir.ai/numbers/6623641c-0d35-4f8a-8fb8-7b844dd80a49apir.ai