Dimittri and his co-founder have launched Sonarly, an AI agent that connects to observability tools like Sentry and Datadog to triage and fix production alerts. The YC W26 startup aims to cut Mean Time To Repair (MTTR) by filtering alert noise and providing root cause analysis. Their tool groups duplicate alerts and reduces the daily alert volume from hundreds to a manageable handful.

Drawing from their own experience running a B2C edtech app that generated up to 50 alerts daily, the team recognized how alert fatigue impacts on-call engineers. While traditional monitoring platforms catch many bugs, they often create overwhelming noise that engineers must manually sift through. Sonarly's approach goes beyond simple aggregation - it builds an internal map of production systems to understand service relationships and provide Claude Code with precise context for fixes.

The platform now integrates with existing tools rather than requiring engineers to adopt new trackers. One user receiving 180 daily alerts saw their workload reduced to just 5 critical issues per day through Sonarly's triage process. The team is actively seeking feedback from engineers about their current alert workflows and offers a generous free tier to encourage testing and input.