Improved: Crossarm damage detection

The crossarm damage model has moved out of beta and now delivers production-ready performance with accuracy exceeding 80%. The model consistently identifies damaged wooden crossarms, enabling utilities to prioritize maintenance with greater confidence and fewer missed issues.

Improved: Pole-top damage detection

The pole-top damage model has graduated from beta to production-level performance with accuracy exceeding 80%. This update improves detection of frayed and damaged pole tops, reducing the likelihood of missing damaged assets during pole evaluations.

Improved: Transformer oil leak and corrosion detection

The former transformer damage model has been refined into two distinct, production-ready classes: transformer oil leaks and transformer corrosion. Each class was trained and validated independently to maintain strong performance while enabling more precise identification and prioritization. This separation allows higher-risk oil leaks to be triaged separately from corrosion findings.

Improved: Asset detection

Upgrades to the core asset detector improve detection reliability for both utility and lighting poles, with particularly strong gains for lighting infrastructure. Once deployed to Edge units, this update will reduce missed asset detections and increase overall dataset completeness across inspections.

In progress: Make-ready and measurement intelligence

We’re actively developing new capabilities to collect highly accurate, engineering-grade measurements directly from a moving vehicle. This work focuses on capturing make-ready-level data at scale and reducing reliance on manual field collection and back-office annotation. The result is a faster path from field capture to approval for attachment service requests. Stay tuned—we’ll be sharing more later this year.

Improved: Security and credential management

Implemented a series of security enhancements to strengthen access controls and credential handling across the platform. Updates include stricter password requirements, centralized secret management with improved rotation, and additional safeguards around credential storage. These improvements support our ongoing SOC 2 compliance efforts.

Improved: Shutdown reliability and stability

Multiple improvements to system shutdown behavior reduce delays, prevent failures, and improve resilience under edge-case conditions. These updates address slow or unavailable network connections, GPS disconnects, vehicle speed fluctuations, and rare shutdown stalls, resulting in more predictable and reliable field operation.

Improved: Edge performance and resource management

Performance updates reduce unnecessary I/O usage, resolve telemetry issues on older units, and fix a camera configuration issue that could lead to excessive memory usage and system resets. These changes improve long-term stability and uptime.