What is Lingdian AI
Lingdian AI Agents is the AI capability hub for delivery and troubleshooting teams. Turn point-table recognition, expert Q&A, device context, and confirmable memory into governed assets — so every commissioning shift starts from what the team already learned.
This section explains what the agents are, what they solve, and how collaboration and memory work — aligned with first-pass connect, remote clarity, and trustworthy data.
How one job becomes the next starting point
ASSET LIFECYCLE · field experience into reusable assets
Capture
Point tables / protocol PDFs / samples, device & alarm context, on-call chat enter the system.
Process
OCR & protocol parse → thing-model draft; assistant recalls knowledge / Skills; sessions compress by layer.
Assetize
Drafts confirmed to store; diagnoses written; memory candidates confirmed and owned by experts.
Assemble
Bind knowledge, Skills, tools per tenant / project / agent; inject only what the task needs.
Reuse & feedback
Next same-type device needs less mapping; next shift fewer calls; results update memory & eval.
Three capabilities you actually care about
Role: less debug, just use
AI sits in the delivery path — not a lone chat box. Point tables, troubleshooting, SCADA assist.
- Point table / protocol → thing-model draft
- Q&A with device & log context
- Human confirm for high-risk actions
Collaboration: the right agent
Platform assistant, device / station agents, system experts & Skills — delegate and assemble across sessions.
- Skills / knowledge by role
- Expert handoff packages
- Same permissions as the user
Memory: evidence → long-term cognition
Sessions compress with provenance; expert memory is confirmable; device / station brains hang on objects.
- Session: raw → summary → scenario
- Expert: candidate → confirm → recall
- Object: device / station memory graph
Boundaries: not a protocol substitute
Protocol gaps stay on hot-pluggable plugins; AI speeds semantics. Memory needs confirm; tenant isolation holds.
- Humans publish & approve
- Tenant / project isolation
- Never put ProductSecret in chat
Start here
| You want… | Open |
|---|---|
| What cost AI removes | AI Role |
| Multi-agent split & handoff | Agent Collaboration |
| How memory grows | Agent Memory |
| Gateway point-table path | MQTT Gateway |