We built SENAITY because we kept watching labs verify results they should not have.
Not because the analysts were careless. Because the volumes were too high, the screens were too dense, and the only safety net against an out-of-range value was someone tired catching it on the last review of the day. A laboratory that verifies five thousand results a week is going to publish a bad one occasionally. The cost of that bad one is what got us writing.
SENAITY is the quality-assurance layer for SENAITE. It scans every analysis as it flows through your lab and flags the values that do not fit your historical pattern, before they reach the verified- and-published step.
What it actually does
Three things, all on top of an existing SENAITE installation.
1. Anomaly detection per analysis, per client
SENAITY uses Isolation Forest models, trained separately for each analysis keyword and each client. Anything that does not fit gets flagged before it can be verified. No manual thresholds, no rule maintenance. Sensitivity is adjustable per analysis: Strict, Normal, or Sensitive.
The model retrains on your verified results, so it stays calibrated to how your lab actually works.
2. A single Kanban view of every sample in flight
Due, Received, To be Verified, Verified, Published. The view auto- refreshes as your SENAITE state changes. Click a sample card and the analysis drawer opens with pending tests, full results, units, uncertainty, and the SENAITY anomaly flags overlaid in line.
Full-text search and dynamic filters on any SENAITE catalog field. Scientific notation rendered properly. The verifier sees the outlier highlighted before they sign anything off.
3. Side-by-side sample comparison
Pick any two or more samples. SENAITY lines them up next to each other, computes %RSD against adjustable thresholds, generates Z- score charts and distribution plots, and shows you where one sample drifts from its peers. Export the comparison as XLSX for documentation and audit trails.
Where it runs
On your own infrastructure, next to your SENAITE. Three small Docker containers behind your existing reverse proxy. Total memory footprint under 500 MB.
Your lab data never leaves your network. No DPA negotiation, no GDPR Article 28 paperwork, no ISO 27001 audit questions on call one. The same procurement path your team already uses for SENAITE itself.
Specifically:
- On-premise install with a single docker-compose stack.
- Anomaly model bootstrapped from your data during install. We pull the last 90 days of verified results from your SENAITE and train the initial model.
- Uses your SENAITE credentials. No new accounts, no extra passwords. SENAITY authenticates against SENAITE itself.
- No data lock-in. Every value lives in SENAITE. SENAITY only reads and overlays. Cancel the subscription, the data stays in SENAITE.
What it costs
Four tiers, from free.
- Community (EUR 0): open-source, self-host. Full feature set including anomaly detection. You run it, you maintain it. The same model the paid tiers run on.
- Lab (EUR 490 per month, or EUR 4,900 per year): standalone Docker license, single lab, on-premise. Includes white-glove install, model bootstrap, a 30-minute training call, and a 14- day follow-up.
- Group (EUR 1,490 per month, or EUR 14,900 per year): multi- site, 3 sites included, EUR 290 per month per extra site. Single invoice for lab networks.
- Enterprise (from EUR 30,000 per year): validated environments (GxP, IVDR), IQ/OQ documentation, signed SLA, frozen versions, source-code escrow on request.
The SENAITY product page has the full breakdown.
How to look at it
Two paths.
- The hosted demo at senaity.com. Click around with our SENAITE, or point it at yours if your SENAITE is reachable from the internet. For evaluation only.
- A 90-minute install on your own infrastructure. We deploy the three containers next to your SENAITE, bootstrap the anomaly model on your last 90 days of verified results, and train one analyst plus the lab manager. For production.
Why we built it ourselves
We are the SENAITE core team. We see every kind of lab data SENAITE runs against, across water, mining, clinical, food and beverage, environmental. The patterns that anomaly detection picks up are not the same in every lab. A water utility’s pH distribution is not a clinical lab’s blood-glucose distribution.
What works is a model that trains on your data, not on the internet’s. Built by people who understand the schema it queries. Shipped as software that runs where the data lives.
That is SENAITY.