Federated pattern review.
Your plant gets smarter when the network does.
Each approved reviewer decision at one plant sharpens the detection model for every other plant in the network — without moving a single customer name, asset tag, or telemetry value across the boundary.
Available with roadmap item #45 — ships when legal model + three or more pilots opt in.
Pattern shapes. Reviewer decisions. Rejection reasons.
- ✓ Anonymized failure-mode shapes (e.g., seal_pressure_low → packing replace × 3 events / 30d)
- ✓ Reviewer decisions (approved / rejected) with rejection reasons for model learning
- ✓ Plant and reviewer identifiers are one-way hashed — no re-identification path
Customer data. Asset tags. Entry bodies.
- ✗ Customer or facility names
- ✗ Asset identifiers, serial numbers, or physical-plant topology
- ✗ Telemetry values or raw reviewer-entry bodies
- ✗ Any data before an explicit opt-in + legal data-sharing agreement
Each approved review improves the federation model — for everyone.
Once federation is live, your reviewer decisions aren't just governance for your plant — they become a moat for your network. Contributing plants see their patterns adopted across the field; every adoption sharpens the detection model they themselves rely on.
Example values for illustration only. Federation model activates after three or more pilots opt in.
Pilot now. Unlock federation when your peers join.
Your first value is local — Sovel detects, captures, and governs on your plant's data alone. Federation is a later-stage unlock, not a day-one requirement.