What Konyx can and cannot show yet, and how its claims are measured.
3 records · checked 2 October 2026 · this domain as JSON · whole catalog as Markdown
Konyx has no published customer results yet. The pilot defaults on the site are process defaults, not outcomes. Results from pilots will be published here with the customer's consent, with the measurement method beside them.
Related: Decide with evidence at day 90 · How savings are measured
Source: konyxai.com — Pilot · Checked 26 September 2026 · Version 1 · proof.customer-results
Every figure in a savings audit is computed in code from the company's own file: the price gap against its own best price for the same item, against the Konyx network and, when linked to a company on the platform, its catalog; cycle time and late deliveries from dates; quote competition from vendor counts; rush and card spend from flags; buyer hours from the step count of the manual process against Konyx's two steps. The model writes the findings from those figures and cannot change them.
Related: The savings audit · Baseline — before any contract · Time saved, RFQ by RFQ
Source: konyx-api — core/savings-analysis.ts · Checked 26 September 2026 · Version 1 · proof.savings-audit-method
Each release runs type checks and lint in every app, a coverage check that every tenant table carries its isolation policy, a live test that a cross-company read is refused, unit and integration tests in the API, and an end-to-end run through the live servers as a signed-in user — admin creating a company, an owner importing a team, a request extracted by the live model, an RFQ sent, a vendor quoting twice, an order issued, isolation refused, and the test companies deleted.
Related: Each company is isolated in the database · The shape of the system
Source: Engineering Action Plan — Build status · Checked 1 October 2026 · Version 2 · proof.engineering-verification