What is Digel?
The industrial AI platform that tells you what needs attention.
Every industrial software tool built in the last 30 years follows the same pattern: you have to know what to ask before you can find anything. A CMMS shows you open work orders if you open the work orders page. A historian shows you a sensor trend if you know which sensor to plot. These are questions apps.
Digel is an answers app. It reads your process continuously and tells you what deserves attention, before you know to ask.
The three layers
Context Graph
Everything Digel knows about your plant lives in an industrial knowledge graph: assets (plants, lines, sections, machines), their relationships, connected sensor tags, uploaded documents, work orders, and maintenance history. This is the memory that makes AI answers reliable instead of generic.
AI Agent
A single AI agent runs throughout the product: in the chat sidebar, in @digel mentions inside work order comments, in triage investigations, and in report generation. The agent does not guess. Every answer it gives is grounded in your context graph, with citations back to the specific asset, sensor reading, or document it used.
What the agent can do:
- Answer operational questions ("Is PEF running? What is the flow rate on Line 2?")
- Investigate problems across sensor data, documents, and maintenance history
- Create and fill in work orders from a plain-language description
- Build dashboards and run Python analysis in a sandboxed environment
- Generate investigation reports from chat
Issues
The Issues view is where the agent's proactive work surfaces. The Triage list shows what the agent thinks needs attention right now, each item written in plain language with the reasoning behind it and one click to act on. You can accept, reassign, dismiss, or argue. Arguing sends the item back to the agent for re-investigation against live data.
Work orders, recurring maintenance templates, and condition-based triggers are all part of the same system. The agent can create a work order from a chat message. A template can fire on a counter value, a measurement threshold, or a time interval.
How the data flows
Your sources (OT sensors, ERP data, PDFs, notes) flow into the knowledge graph. The knowledge graph feeds the AI agent. The agent feeds the Triage list and responds to your questions. Every layer references the same underlying model.