Lighthouse Projects · Energy
Operations copilot for EV charging networks
Questions about stations, sessions and invoices, answered in natural language.
EV network
15 stations
Anomalies
4
▌ Copilot · Question
“Which stations have the most failed sessions today?”
Problem
Data scattered across EV platform, billing, CRM, SCADA. Ops teams jumping between systems to answer simple questions about availability, failed sessions and billing errors.
Solution
AI agent with semantic model over stations, sessions, customers and invoices. Operators ask in natural language. Instant diagnosis. Operational action suggestions (open ticket, adjust tariff, contact customer).
KPIs
- Average time to operational answer
- Average network uptime
- Reduction in duplicate billing tickets
- Time to identify critical stations
How it is measured
Every KPI with a baseline and a target.
The numbers for this project come from your data. We measure the baseline during framing, agree the target with you before building and, in production, measure the same KPIs the same way.
- 01BaselineMeasured during framing, with the data of the process as it runs today.
- 02TargetAgreed with the case owner before building, starting from the baseline.
- 03In productionThe same KPIs, measured the same way, compared with the baseline.
- Average time to operational answer
- From question to answer, on a sample of real questions recorded before and after.
- Average network uptime
- Time each charge point was available, from the statuses the charging management platform records.
- Reduction in duplicate billing tickets
- Tickets classified as duplicate billing in the CRM or support tool, per month.
- Time to identify critical stations
- From the first alarm or failed session until the station is flagged to the operations team.
Before starting
Prerequisites and risks.
Prerequisites
- Read access to the charging management platform, billing and the CRM.
- Agreed definitions of an available station, a failed session and an invoice in error.
- Operators to test everyday questions during the pilot.
Risks and how we handle them
Answers with wrong numbers
The agent answers over the semantic model, with the agreed definitions, and shows the query it ran. A question outside the model gets no answer rather than an estimate.
Actions on the network
Opening a ticket, adjusting a tariff or contacting a customer are suggestions that an operator confirms. Nothing changes on the network without that confirmation.
Want to see this in your operation?
90 minutes with you and your team. We leave with 3 concrete opportunities and estimated payback.