Superlog finds the production bugs your monitoring never sees. It scans every log line, spots error patterns no alert is watching, investigates them 24/7 and proposes the fix. It works on top of Datadog, Sentry, Grafana and OpenTelemetry. Used by 100+ startups.
This is the 3rd launch from superlog. View more

superlog
Launched this week
Your monitoring only alerts on what you told it to watch, so customers still report bugs. Superlog finds the errors nobody set an alert for, investigates them 24/7 and hands you a fix. 100+ startups use it; 87% of our fixes get merged.





Launch tags:Developer Tools
Launch Team

AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted



Hey Product Hunt 👋 Nicolò here, co-founder of Superlog (YC) with Arseniy (ex-Datadog).
Every team we meet already has Datadog, Sentry or Grafana. They still hear about bugs from customers first. Monitoring only alerts on what someone expected to break.
Superlog doesn't replace your monitoring. It works on top of it and finds everything else:
Finds what your stack misses. Our algorithms scan every log line and spot new error patterns no other tool catches.
Investigates 24/7, on its own. It finds the root cause in your code and skips what you already track.
Proposes the fix. You fix it your way, with your own agent and the full context, or we open the PR.
100+ startups use it today, including Plato, Alara, Nautilus and Trellis. They merge 87% of the fixes we suggest. We're backed by people from Google, Datadog, OpenAI and Slack.
Connect one service and see what you've been missing. Code PRODUCTHUNT gives 50% off Pro for 3 months. What's the worst bug a customer found before you did?
@nicolo_magnante pre-launch so it's testers for now, but yes. one sent us a list of things our own checks never flagged. nothing was erroring, it just felt wrong. curious if log patterns can catch that kind
When it suggests new logs or metrics to add, does that come through as a PR too, or just as a note in Slack?
Like the findings based pricing - very clear cost -> value mapping. However, curious about
How do you benchmark yourself against claude/codex loop raising error patterns?
Is there any limit on amount of logs one can send? Isn't work involved on your end to find error patterns proportional to amount of logs being sent?
The part that stuck with me is handing the bug's context to my own coding agent instead of a black-box fixer — that's how I actually work, I want the fix to happen in my repo with the full context attached. As a solo dev I don't run Datadog or Sentry at all though, so I'm curious: does it need a full observability stack to plug into, or can it work off plain server logs?
Congrats on the launch!
Our test runs go at the dev BE, and some specs walk failure paths on purpose. in the logs that reads the same as someone actually stuck.
How does the user impact step handle traffic thats not a person?
How do you folks understand what's a production issue? How's it better than Datadog/Sentry/Grafana?
@alesia_cherniavskaia great question!
first of all, we fingerprint every log/trace/metric pattern and find good candidate issues to investigate
for every candidate, we will check if
a) some real users are blocked from using the product
b) you already know about this issue (via alerts / github prs / linear tickets)