Superlog bets observability should fix bugs itself, not page humans

Superlog is a two-person San Francisco startup from Y Combinator's Spring 2026 batch, founded by Nicolò Magnante, an ex-BCG consultant who had already scaled two startups to millions in annual recurring revenue, and Arseniy Shishaev, who built high-performance data systems at Datadog and spent nights being paged for production incidents. Their bet is that observability's problem is no longer visibility but the human loop around it: setup, alert fatigue, triage, and fixes.

The product is deliberately described as 'the observability tool you're not supposed to open.' A wizard scans a repository and instruments it with native OpenTelemetry, then re-runs daily so logs, alerts, and dashboards keep pace with code that ships. Errors are fingerprinted and grouped into incidents with severity and impact shown upfront; an agent investigates with full context and either produces a tested fix or tags the engineers who own the system.

Superlog launched publicly on Hacker News on May 19, 2026, drawing 74 points and 49 comments, then released its core under an Apache 2.0 license in June and started Superlog Cloud with a free tier, usage-based pricing, and Pro and Max plans. EveryDev.ai's June 2026 profile counted 640 GitHub stars on the open-source repository, which contains the web app, API, OTLP intake proxy, worker processes, and database schema.

By late August 2026 the company remained a two-person team with no disclosed revenue, customers, or institutional round beyond YC's standard investment. The open question its own pitch poses: whether engineers will trust an AI agent to edit their code path in production enough to let a monitoring tool close the loop, or whether the merge button stays a human one for a long time.