Building a Local Industrial AI Agent
An architecture discussion: PLCs, meters and sensors through a gateway to edge data, a local model and an agent that answers questions about the plant — and the limits to respect.
Series 04 — Edge, IIoT and AI
About this webinar
Running a model on site raises practical questions long before it raises clever ones: which data, at what rate, stored where, and who is allowed to act on the answer. This session is a working session on that architecture, presented as a direction rather than a finished product.
What you will learn
- Collecting from PLCs, meters and sensors into a consistent edge dataset
- Local history and retention: what to keep on site and for how long
- Interfaces an agent can use: HTTP API, MQTT, OPC UA and a semantic naming layer
- Read-only by design: why an agent should explain before it is allowed to act
- Where this stands at MILESGO today: pilots, roadmap items, and what is not claimed
Who should attend
- Senior automation engineers
- Software developers working with industrial data
- IT and OT architects
- Digitalisation and innovation teams
Agenda
- 10 min — Where industrial data sits today, and why it is hard to use
- 15 min — The data path: device, gateway, edge, API
- 10 min — What is possible today, what is still a pilot
- 5 min — Security and data ownership questions to settle first
- 5 min — Q&A with the MILESGO technical team
Speaker
MILESGO Technical Academy — industrial connectivity and IIoT engineers, with partner speakers for the industry sessions.
How to attend
Sessions run live per region and on request for a single company or project team, in English or Chinese. Contact MILESGO to be put on the list for the next session, to request this webinar for your team, or to ask for the recording once a session has been held.