24/7
on-demand support for distribution engineers, answered by voice from the utility's own network model.
A voice agent that sits on top of the distribution digital twin, so engineers in the field and in the office can ask the network a question and hear the answer, with the source it came from.
- Industry
- Utilities
- Segment
- Electric distribution
- Stage
- Build Agents
- Status
- Delivered
Digital Twin Voice Assistant
The findings in four sentences.
- 01
The bottleneck was never the model. It was finding the one approved document that answers the question.
- 02
Engineers trusted the agent once every answer showed its source. Before that, they re-checked everything by hand.
- 03
Voice mattered most for field crews. Office engineers preferred the same agent in chat.
- 04
A short list of questions the agent must never answer on its own did more for adoption than any accuracy metric.
Client problem
What wasn't working.
- Distribution engineers spend a large part of the day hunting for answers across GIS, the network model, standards manuals and old study reports.
- Knowledge about why the network is built the way it is sits with a shrinking group of senior engineers.
- Field crews can't stop to open three desktop applications to confirm a rating or a switching configuration.
Dviz solution
What we built.
- Built a voice agent on Sorgi that reads the digital twin, the standards library and approved study reports in place, without copying the data out.
- Every answer cites the asset, document or model element it came from, so engineers can check it before acting.
- Questions that touch switching or protection settings are routed to a human reviewer instead of being answered automatically.
- Delivered as a repeatable agent pattern the utility can point at new data sources.
Engineer
What's the rating on feeder 12-F4 past the Elm St recloser?
Agent
336 ACSR, rated 530 A summer normal. Source: network model, segment F4-118.
A working sketch of the system, with placeholder data.
Delivered on Sorgi
How the platform carried it.
- Connectors read GIS, network model and document stores in place.
- Human-in-the-loop checkpoints for switching and protection questions.
- Full audit trail of every question, answer and source.
Outcomes
24/7
Answers available to field and office teams
100%
Answers tied to a source in the twin or standards
Next use case
AI-Assisted Spec Generation
Have a problem like this one?
A backlog nobody can clear, a pilot that never left the lab, a process that only three people understand. Tell us what it is, and we'll tell you honestly whether AI is the answer.