Zero-config anomaly detection
Stop tuning a threshold for every single device.
BIoTan compares each asset against its peers — so shared swings like weather, load, and seasonality cancel out, and only the true outliers surface. No per-device setup.
Validated on 7 independent public datasets — including NASA's C-MAPSS turbofan fleet, where peer-z rose above 2σ before failure for 99 of 100 engines.
See a live example report — no signup →How it works
- 1
Upload your history
A CSV of past readings from your fleet — device_id, timestamp, value.
- 2
We compare peers
BIoTan ranks each asset against similar ones and tracks how it drifts over time.
- 3
See the lead time
For known failures, we show how many days earlier BIoTan would have flagged the asset.
Why peer-relative?
Absolute thresholds drown in false alarms the moment conditions change. BIoTan asks a different question: is this asset behaving unlike its peers right now? Common-mode variation is removed automatically, so you configure nothing and still catch what matters.
- No per-device thresholds to set or maintain
- Common-mode noise (weather, load, season) cancels out
- Flags assets drifting away from their peer group
What earlier warning looks like
On NASA's C-MAPSS FD001 fleet — 100 engines run to failure — common-mode removal lifted peer-z above 2σ before failure for 99 of 100 engines, and the conservative zero-config gate confirmed the fastest-degrading engines a median of ~11 cycles ahead. Upload your own history to see the same view on your fleet.
Lead time is measured in hindsight, so treat it as an optimistic upper bound — not a guarantee of future warning.
What BIoTan is — and isn't
BIoTan is a risk-prioritization and degradation-tracking tool, not a crystal ball. Failures with no precursor signal can't be predicted by any tool, and we won't pretend otherwise.
- If the signal is weak or the data is short, the report says “no clear precursor” — we don't invent false positives.
- It surfaces and ranks degradation; it does not promise to catch every failure.
- Lead-time figures are optimistic upper bounds from after-the-fact analysis.
Open core
Free, source-available core you can self-host. Paid enterprise tier for real-time operation and connectors. Your data stays local — no telemetry by default.
Core (Free)
Batch analysis, self-hostable, full algorithm access. Run on your infrastructure with your data.
Enterprise
Real-time streaming, SCADA/historian connectors, SLA support. Contact us for pricing.
Your data, handled honestly
- Uploads are used only to generate your report, then deleted immediately.
- Nothing is stored, logged, or shared with third parties.
- Open core — you can self-host the engine yourself.