How Sunrun Powers America's Leading Solar Operations

About Sunrun

Sunrun is America's leading residential solar company, helping homeowners transition to clean, affordable energy.

As a publicly traded company serving customers nationwide, Sunrun operates at the intersection of renewable energy and cutting-edge technology.

Scaling AI

As Sunrun's AI team launched new projects, their automations and applications relied heavily on LLMs, so they needed infrastructure that could scale without slowing down development.

Most importantly, the team needed caching that worked out of the box. They also needed critical visibility into their AI operations to react quickly when things went wrong.

The observability and alerts features are really important to us. If any of our providers are having an outage, it's just important for us to be aware of that immediately.

Helicone since Day One

Sunrun adopted Helicone at the very beginning of their AI journey.

For Sunrun's AI team, three capabilities stand out as essential:

1. Built-in Caching

Helicone's caching layer worked immediately without requiring custom infrastructure.

It was great how [caching] was built in. It just worked!

2. Real-Time Alerts and Observability

When provider outages happen, the team knows instantly. The ability to debug requests and see exactly what's happening, meeting the team in Slack where they can react quickly, is crucial for the maintenance of their reliable customer-facing applications.

3. Property-Based Analytics

Recently, the team started leveraging Helicone's properties to track special use cases and assess performance, cost, and latency across their product.

This granular visibility helps the team identify performance bottlenecks and optimize their applications for speed—critical support when serving residential solar customers who expect responsive, reliable service.

The properties are really nice for being able to see what is costing us the most or taking the longest time. Mostly latency is what we care about for most of these use cases. So being able to see what property causes more latency than others is quite nice.

The Results

By integrating to Helicone from day one, Sunrun built their AI operations on a foundation of observability and reliability:

• Zero setup overhead: Caching and observability worked immediately without engineering effort
• Instant issue detection: Real-time alerts so the team knows when providers experience outages
• Latency optimization: Property-based analytics revealed which operations need performance tuning
• Focus on what matters: The team can build features instead of maintaining custom-built infrastructure

The caching, the observability, and the alerts are mostly what matters to us.

Looking Ahead

As Sunrun continues expanding its AI-powered operations across residential solar installations nationwide, integrating to Helicone means they can focus on shipping features rather than maintaining infrastructure.

By leveraging caching, observability, and alerting since the very beginning, the engineering team has been able to support American homeowners in harnessing the power of clean, affordable solar energy at scale.