In the product
What it looks like
Screens from the live platform, demo tenant. Click any screen to open it full size.
Drop a manufacturer datasheet and get a typed specification and a five-facet measurement form. Drop an old certificate PDF and get a template. Drop your old system's export and get an approved column mapping. Every AI output is a draft for a qualified person, and the provenance ledger records the model, the prompt hash, the output, the citations and who approved it.
The extractor reads the manual, proposes tolerances with verification tiers, and stores the document, revision and clause. Cal Builder drafts points, tools, rounding, acceptance and environment. The uncertainty study builder proposes contributors from published guides. Out-of-tolerance investigations and CAPAs are drafted from reverse traceability. The certificate states when a specification was AI-assisted — and only when the origin is the manufacturer.
Screens from the live platform, demo tenant. Click any screen to open it full size.
The datasheet is read; tolerances are proposed with verification tiers and their source clause.
A review queue: approve, dismiss, override per lab.
Cal Builder drafts the measurement form; the uncertainty builder drafts the study; the report builder rebuilds an old layout.
A qualified person signs. The ledger records who, what model, what prompt, what output, edited or not.
The certificate states AI assistance when the specification came from the manufacturer with AI help — and stays silent otherwise.
A quality system is defined as much by what it will not accept. These are enforced in the platform, not in a procedure.
Nothing is signed by a model. Every output is a draft until a person approves it.
The flag is cleared; the certificate says nothing about AI.
Drop a manufacturer datasheet and get a typed specification and a five-facet measurement form. Drop an old certificate PDF and get a template. Drop your old system's export and get an approved column mapping. Every AI output is a draft for a qualified person, and the provenance ledger records the model, the prompt hash, the output, the citations and who approved it.
A small illustration of the extractor's first job: recognise the tolerance type, split its terms, and note where it came from. The real one reads the whole document and proposes verification tiers.
Extraction with provenance is where they stop — and the reviewer sees the clause it came from.
An assessor can see what was drafted, by which model, and who approved it. The certificate declares it.
Calibration & uncertaintyCalibration & uncertaintyThe verdict is computed, not typed.Every calibration is a numbered event that carries its own points, readings, reference standards, environment and decision rule — and evalua…
Quality management · QMSQuality management (QMS)A deviation the system can explain.CAPA, NCR, deviations, complaints, controlled documents, training, audits, risks, change control, vendors, methods, proficiency testing — on…
Enterprise asset managementEnterprise asset management (EAM)Every asset. Every part. Every document.A real hierarchy — customer → site → building → room → bench, with parent and component systems — typed specifications with provenance, cust…
Integrations · API · automationIntegrations, API & automationConfigured, never hard-coded.A visual integration builder with a ten-system catalog plus generic REST and CSV, AI-proposed field mapping with a dry run, signed webhooks …The tour runs the working platform, not a mockup. Pilot on your own instruments with your own imported history — we set it up with you, and fixes ship in hours and days, not release cycles.