Kaupify has its own accounting ledger, invoices, bills and financial reports. Here you can see the checks behind selected workflows, the conditions used and what still needs validation.
These results describe specific exercises. Test suites can overlap; their counts are not added together as a measure of complete product coverage.
Documented exercise
A month of accounting workflows
66 workflow checks passed using fictional companies and records.
What this covers
Selected invoice, bill, payment, manual CSV bank-reconciliation, period-close and company-isolation scenarios exercised through the application routes.
Limits of this result
Synthetic records and controlled provider behavior, with test identities enrolled for current MFA requirements. This is a selected workflow exercise, not validation of every feature, bank or country setup.
Documented exercise
Accounting integrity and controls
Three additional accounting selfcheck suites passed: integrity, controls and hardening regression.
What this covers
Selected ledger integrity, posting controls and regression scenarios. The separate performance dataset also passed 16 financial and database checks.
Limits of this result
Internal engineering checks on a fixed source revision. The hardening race scenario does not establish multi-user capacity. A complete product qualification has not passed.
Documented exercise
Access and security regression checks
724 targeted cases passed, with one documented skip and no failures or errors in that run.
What this covers
Selected authentication, tenant separation, support throttling and release safeguards in the 17 October security release.
Limits of this result
Dated regression evidence, not a fresh full-product security audit. The skipped case concerned a reverse-proxy configuration outside that release. No independent penetration test or security certification is claimed.
Documented exercise
Encrypted backup recovery exercise
An isolated exercise recovered 20 databases and 308 runtime files from an independently downloaded encrypted backup.
What this covers
Verified backup download, ciphertext matching, extraction and removal of temporary plaintext after the exercise.
Limits of this result
The exercise did not boot the restored application or rebuild Linux services, private configuration, proxy, TLS or DNS. It does not establish a complete disaster-recovery time.
Documented exercise
Public website release verification
The accounting-information release passed 89 native Linux checks and its desktop, phone and public HTTPS checks.
What this covers
That release covered public accounting descriptions, discovery files, page metadata and preservation of private-access boundaries.
Limits of this result
These checks covered that release, not complete accounting or security qualification. Counts overlap with other suites and are not added together.
02 / Performance evidence
The workload behind the numbers.
Operation timings depend on the work, data and environment. A measured result applies to those conditions and does not establish a capacity guarantee for every customer or workload.
Tested on
2026-10-22
Test records
One fictional company: 1,000 sales invoices, 1,000 supplier bills, 1,000 allocated payments and 105 manual-journal reversal pairs; 3,210 journals and 8,420 entries in total.
Environment
Linux x86_64; Python 3.12.15; SQLite 3.34.1. One sequential worker on the application VPS, isolated from customer data; WAL mode and synchronous=NORMAL.
Measurement scope
Backend posting and real report queries only. Reports include opening and closing a SQLite connection. Warm cache; five initial warm-ups excluded. Each posting row has 100 samples; each report row has 50. The measured financial engine is unchanged in this page release.
Recorded operation times in milliseconds
Measured work
Samples
Median
95th percentile
Post a sales invoice
100
0.224929 ms
0.513496 ms
Post a supplier bill
100
0.242115 ms
0.418119 ms
Record and allocate a customer payment
100
0.149865 ms
0.316483 ms
Record and allocate a supplier payment
100
0.160666 ms
0.31377 ms
Post a manual journal
100
0.060522 ms
0.093553 ms
Reverse a manual journal
100
0.237539 ms
0.516559 ms
Query the trial balance
50
3.478143 ms
4.304222 ms
Calculate profit and loss
50
3.27 ms
3.483477 ms
Calculate the balance sheet
50
3.093856 ms
3.260847 ms
Median is the middle recorded time. The 95th percentile is the time at or below which 95% of the recorded samples fell.
How to interpret the measurements
HTTP requests, browser rendering and network latency are excluded. These are operation timings, not end-to-end page-load times.
The database grows during posting phases; reports use the final dataset. One sequential writer was measured, so no concurrent-user capacity, uptime or response-time guarantee is established.
The VPS also hosts other services and the test has resource and audit limits. Results describe this run and workload; different customer data and workloads can differ.
Performance checks do not establish complete feature coverage, integration approval, country compliance or that Kaupify is safer than another product.
03 / Clear boundaries
Useful evidence. A continuing process.
Complete product qualification is in progress. Selected passing checks do not mean every feature, country setup, integration or failure scenario has been validated.
Engineering checks
The results describe internal verification. They are not an independent penetration test, security certification or certification of the complete product.
Your accounting setup
Correct accounting reports depend on the records, opening balances, country settings and accounting treatment. Review your setup and filing requirements with your accountant.
Your connected services
Bank connections and payment services need their own supported provider, configuration and approval. Controlled tests do not establish that every provider is ready for production.
Questions about the evidence
These checks do not establish that Kaupify is safer than another product. Ask us about a workflow your business needs, and we can explain the available evidence and remaining checks.
Evidence you can inspect
Keep the detail with the claim.
The public evidence summary records the dated scope and limitations shown here. It contains no company records or credentials.