Your plant's raw signals, structured in real time.
The edge-native AI industrial platform.
An AI layer that speaks your factory's language.
edgentia's models run next to the machines — typing raw signals, answering questions in plain language, and proposing actions that a human approves. Local models, your data, full context.
Signal inference
An ML model types every incoming tag and places it in the ISA-95 hierarchy — turning a cryptic MQTT firehose into a structured asset graph, with no manual tag-mapping.
torque → asset:lyon-04.stamp-b · 0.98Ask in plain language
Query the live graph in plain language — "which cells on line 4 are running hot?" — and get answers grounded in real-time metrics and asset history, not a generic chatbot.
chat.queried · grounded in live dataAgents with a human in the loop
Agents surface anomalies and propose actions through MCP tools — but nothing runs until an operator approves. Every proposal and approval is logged with full context.
proposal.approved · operator: a.harkous-
ISA-95 scoped access
The AI layer respects the same scope every person does.
-
Full audit log
Every query, approval and tool call recorded — who, what, when.
-
Your models, your hardware
Claude, Llama or Azure OpenAI — on your own edge box.
A shared workspace for the whole shift.
Operations is a team sport. edgentia gives every team a live workspace where context, dashboards and notes stay tied to the asset — so the next shift picks up exactly where the last one left off.
-
Team workspaces
Spin up a workspace per line or team, scoped to its branch of the ISA-95 graph. Comments and shift notes stay attached to the asset — no scattered emails.
-
Drag-and-drop dashboards
Build a live view by dragging control modules from the ISA-95 catalogue onto the grid. Layouts snap into place and persist for everyone on the team.
-
Build with AI
Describe what you want to see — "uptime and jam count for line 4 this week" — and the AI builder assembles the dashboard, picks the tiles and wires the queries.
A founder-led startup, building from the edge up.
edgentia is founded and built by Ahmad Harkous, an IoT and edge-computing engineer. The name says what it does — edge + ia (French for AI), combined into one word.
The conviction is simple: industrial data should be structured where it's born, on the plant floor — not shipped wholesale to a cloud to be untangled later. One runtime, close to the machines, turning raw topics into a graph teams can actually use.
See your plant as one asset graph.
The edge-native AI industrial platform.
Need a hand? Reach us at [email protected]