Hey YC! We're Nicolò and Arseniy, co-founders of Superlog (superlog.sh).

TL;DR: Superlog is the observability tool you're not supposed to open. A wizard scans your repo and instruments it with OpenTelemetry, and re-runs daily to keep your logs, alerts, and dashboards in sync with your code. When something breaks, an agent investigates with full context and posts one mergeable PR per incident into Slack.

https://www.youtube.com/watch?v=AeX3pnHCMHg

The Problem

Arseniy spent years at Datadog, getting paged during the night to debug production incidents. In our first startups, we tried Sentry, Datadog, Grafana, Dash0, and nothing was good enough.

• Setup never ends. Proper telemetry takes days of manual work and breaks the moment you ship new features. Half the dashboards end up empty or outdated.
• Alert fatigue is constant. Sentry floods Slack with duplicates and no context. The Saturday #ops ping is the worst feeling of the week.
• The bill climbs. Datadog invoices keep growing while engineers still spend hours configuring dashboards, chasing three AWS metrics that disagree, and copy-pasting stack traces into Claude at 11pm.

Our Solution

1. Telemetry that sets itself up. Tell your agent:

Run npx skills add superloglabs/skills --all and use the skills to install Superlog in this project

This skill instruments your code with native OTel SDKs: proper semantic conventions, service and environment tagging. We cover all failure modes, endpoint performance, usage per tenant, and LLM/upstream cost (by callsite, tenant, and model). We run observability check-ins that keep adding logs, alerts, and dashboards as your code grows.

2. An agent that investigates and ships PRs. Errors get prioritised and grouped into incidents. The agent investigates with full context (logs, traces, metrics, recent deploys) and either produces a short tested PR, or posts its findings and pulls in the engineers who own the system (based on docs, prior investigations, Slack threads). One mergeable PR per incident in Slack. Merge it, ignore it, or open it as a Claude Code/Codex session and tweak.

3. No alert fatigue. Our custom eval setup makes sure incident summaries in your #ops channel are dense and correct, with severity and impact upfront. Every LLM-enhanced metric ships with a confidence score, so wrong guesses don't get boosted.

The telemetry you get is vendor-neutral: you keep all the OTel logs, metrics, and traces we install, even if you choose to switch providers.

Where We're Going

Every engineering team in the world burns nights and weekends debugging production. We want to delete that loop. Observability should install itself, alert intelligently, and fix what it can.

About Us

• Nicolò (CEO) — ex-BCG, math olympiad winner, scaled two startups to millions in ARR, with international enterprise contracts at Bluco (McDonald’s, Volkswagen, Dufry, etc).
• Arseniy (CTO) — coding since 12yo, won school olympiads, worked on high-performance data systems at Datadog and built engineering teams from scratch at Bluco.

Our Ask

We'd love intros to:

• Engineering leaders drowning in Sentry alerts or Datadog bills
• Startups that rolled their own observability and want to throw it away

Try it: superlog.sh
Email: nicolo@superlog.sh