Control Seat V0.1.0
V0 shipped dashboards. V0.1.0 ships the other half of any real operational stack: Flows, a visual rule engine for wiring up real-time alerts, automations, and anomaly detection — no code required for most tasks. Drag triggers, logic, and actions onto a canvas, connect them with arrows, and publish. Your rules run on the gateway, not in a browser tab.
What's in V0.1.0
Flows: visual rule engine
Build rules as a graph of nodes connected by edges. Messages enter through a trigger, flow through logic and transformations, and end at an action. Common things you can build in a few minutes:
- Email ops when pump temperature stays above 180°F for 30 seconds
- Write a daily setpoint at 6am based on yesterday's production total
- POST every active alarm to a Slack webhook
- Catch slow sensor drift before it crosses a fixed threshold
See the Flows documentation for the full walkthrough.
Triggers for any event
Five trigger types cover the common entry points:
- Tag Change — fires when a tag value changes (optionally filter out no-op writes)
- Manual Run — fires when you click Run Now; great for testing and one-off jobs
- Alarm Event — fires on any alarm's active / inactive / ack transition
- Schedule — cron-based, in UTC
- Webhook — exposes a URL that fires the flow when POSTed
A growing node library
Beyond triggers: Read / Write Tag and Historian Query for tag I/O; Expression, Branch, Switch, Threshold, Change, Join, Delay, Throttle, and Debounce for logic and timing; HTTP Request and Notify for side effects; and Subflow for reusing a piece of logic across many flows. Every node has inline help and a typed config schema.
Anomaly detection without fixed thresholds
Three new detectors go beyond static setpoints:
- Anomaly (Z-Score) — rolling or EWMA z-score on a numeric tag. Ideal for sensors whose "normal" range drifts with ambient conditions.
- Trend Alarm — linear-slope, EWMA, or CUSUM drift detection for slow degradation the simple thresholds miss ("bearing temp rising 5°F/hour").
- ML Predict — escape hatch for heavy ML. POSTs a sample or rolling window to an external service (AWS Lookout for Equipment, Azure Anomaly Detector, a custom endpoint) and routes the score through the rest of your flow.
Wire any of them into Notify and you have end-to-end anomaly alerting in three nodes.
AI flow assistant
Describe what you want in plain English and the AI wires it up. "When pump temp goes above 180 for more than 30 seconds, email ops@example.com" — the assistant searches your tags, picks the right nodes, configures them, and lays out the canvas. You review the proposed change before it's committed.
Live payload inspection
Click any edge on the canvas to see the last message that flowed through it — pretty-printed JSON, plus a list of copyable reference paths like msg.payload.items[0].value you can paste straight into another node. Updates live while the flow is running.
Manual run with persistent test payloads
Every Manual Run trigger has a JSON payload editor and a Run Now button in its inspector. Test payloads persist per-node across reloads and sessions, so you can keep a library of scenarios pinned to different triggers.
Draft and publish workflow
Drafts save automatically as you edit. A background preview executor runs your draft so the live payload pills on edges reflect in-progress work. When you're ready, Publish snapshots the draft and the gateway-worker starts firing real triggers within a second.
Safer external calls
HTTP Request and ML Predict now block calls to localhost, private IP ranges, and cloud metadata endpoints by default — a baseline SSRF guardrail for flows that take templated URLs from user-controlled data.
Catalog, prompts, and docs stay in sync
The editor, the runtime, and the AI assistant all share one node catalog. Add a node's description once and it shows up in the inspector, the AI's system prompt, and the search_node_types tool. Less drift between what the docs say and what the editor does.
We're shipping the rest of V0.1 as a steady drumbeat of improvements from here. Read the Flows docs or get in touch for a demo.