A heatwave shifts baseline temperatures. A factory line change alters vibration norms. You need new alert thresholds on 10,000 edge collectors today — not after an OTA firmware cycle.
Kiponos.io connects Python edge agents over WebSocket. Thresholds live in local memory; ops pushes delta updates from the hub.
Edge collector hot loop
def evaluate(sensor_id: str, reading: float, kiponos) -> str:
cfg = kiponos.path("sensors", sensor_id)
high = cfg.get_float("alert_high")
low = cfg.get_float("alert_low")
if reading > high:
return "ALERT_HIGH"
if reading < low:
return "ALERT_LOW"
return "OK"
Called on every sample batch. Reads must be local — no cloud round-trip per reading.
Fleet config tree
sensors/
boiler_12/
alert_high: 92.5
alert_low: 18.0
sample_interval_ms: 1000
conveyor_7/
alert_high: 4.2
alert_low: 0.1
sample_interval_ms: 500
defaults/
alert_high: 100.0
sample_interval_ms: 2000
Ops workflows
| Task | Kiponos action |
|---|---|
| Heat wave | Raise alert_high for outdoor sensors |
| New equipment | Clone folder, tune thresholds live |
| Reduce uplink cost | Increase sample_interval_ms fleet-wide |
| Incident | Tighten thresholds on affected segment |
Each edge agent runs one Kiponos SDK connection. Dashboard change → delta patch → all connected agents update in memory.
Why not poll a central API?
Polling adds latency, battery cost on edge, and failure modes when the network blips. Kiponos keeps last known good values locally and syncs deltas when connected.
Start at kiponos.io. Resources: github.com/kiponos-io/kiponos-io
Kiponos.io — real-time config for Python. Calibrate the fleet while it runs.
Top comments (0)