In early 2023, Voltway Energy had 5,000 EV chargers across 8 states. By the end of 2024, they had 50,000 chargers across 40 states. That's a 10x scaling event—accomplished without a single new operations hire.
How? AI agents handle everything from monitoring to incident response.
The Challenge#
EV charging is a brutal operations problem. Chargers fail in dozens of ways—payment terminal issues, network outages, hardware faults, vandalism, vehicle compatibility issues, software bugs. And when one fails, frustrated drivers leave bad reviews instantly.
At 5,000 chargers, Voltway's 4-person ops team was already maxed out. They were:
- Manually triaging 200+ alerts per day
- Spending nights and weekends on PagerDuty
- Missing failures because alerts weren't actionable
- Losing $40K/month in lost charging sessions
The plan was to grow to 50,000 chargers in 18 months. At current operational efficiency, that would require 40+ ops engineers. The CFO laughed when we proposed the headcount. Something had to change.
The team evaluated traditional ITSM tools, custom-built scripts, and even outsourcing to a 24/7 NOC team. Nothing scaled.
The Solution#
Voltway's CTO had used Evolv Ai at a previous company. He proposed building "a digital ops team" powered by AI agents.
The Monitoring Agent
The first agent reads telemetry from every charger every 30 seconds—voltage, temperature, network signal, payment terminal status, and 40+ other signals. When something looks off, it:
- Checks if the issue is known (matches a pattern from past incidents)
- Classifies severity (P0 if drivers are blocked, P3 if it's preventative)
- Triggers self-healing when possible (network resets, payment reauths)
- Pages humans only when judgment is needed
The Field Dispatch Agent
When a physical visit is required, a second agent:
- Identifies the closest field technician
- Checks their schedule and parts inventory
- Creates the work order with full context
- Sends GPS-optimized routing
- Tracks completion in real-time
The Predictive Maintenance Agent
The third agent analyzes long-term patterns to predict failures before they happen. It now schedules ~40% of maintenance proactively, before drivers ever experience an issue.
“
Our ops team used to be reactive firefighters. Now they spend their days on strategic projects—negotiating with utility companies, optimizing our network, planning expansion. The agents handle the firefighting.
”The Architecture#
Voltway runs Evolv Ai agents inside their VPC, with direct connections to:
- Telemetry pipeline (Snowflake + Kafka)
- Incident response (PagerDuty + Linear)
- Field operations (custom dispatch system + Twilio)
- Customer comms (proactive SMS to affected drivers)
All agents are built using the Evolv Ai SDK with custom tools for Voltway's specific systems. The engineering team owns the agent logic, while the ops team configures behaviors through a simple admin UI.
The Results#
The numbers tell the story:
99.7% network-wide uptime with a 4-person ops team managing 50,000 chargers. That's 12,500 chargers per ops engineer—an industry-leading ratio.
Operational Impact
- Network uptime: 96.2% → 99.7%
- Mean time to resolution (MTTR): 47 min → 5 min
- Alerts requiring human action: 200/day → 8/day
- Customer complaints: Down 84%
- Revenue from improved uptime: +$2.8M annually
Team Impact
Just as important as the numbers: nobody on the ops team has been paged at 2 AM in 6 months. The agents resolve overnight incidents autonomously, with humans reviewing the next morning.
“
I sleep at night now. That sounds small but it's life-changing. The agents handle the boring stuff so well that when I do get paged, it's always something genuinely interesting.
”Business Impact
Voltway's investors took notice:
- Operating margin improved from 8% to 23%
- Series C valuation doubled, in part due to operational efficiency
- Driver NPS scores jumped from 31 to 68
What's Next#
Voltway plans to triple their network to 150,000 chargers by 2027—and they expect to do it with an ops team of fewer than 10 people.
Future initiatives include:
- Grid optimization agents that schedule charging based on renewable energy availability
- Dynamic pricing agents that adjust per-station pricing based on demand and grid conditions
- Compliance agents that automate utility company reporting across all 40 states
"The future of operations isn't more people. It's better
