superlog
Make your product bug-free
Start new thread

Superlog Responder - Free, open-source AI bug-fixing agent

by
•
Responder is an AI bug-fixing agent that plugs into the Sentry or Datadog Slack channel you already run. One-click synch, no new telemetry to install. On every alert it investigates with full context, filters out the noise, and for real issues replies right in the thread with the root cause, the evidence, and a mergeable PR. Prompts, memory, repo access, and escalation rules are fully customizable, so you're building your own debugging agent, not renting ours.

Add a comment

Replies

Best
Maker
📌
Hey ProductHunt! Nicolò here, co-founder of Superlog. A lot of you started using Superlog after our first launch, and honestly the feedback shaped this one. The thing we kept hearing: "I already have telemetry set up, my alerts already land in Slack. I don't want another dashboard to click through. I just want the bug fixed." Fair. So we built exactly that. Superlog Responder lives in the ops channel where your alerts already show up. When something breaks, it doesn't ping you to go investigate, it investigates itself, using your Datadog, your Sentry, your Notion, your repo, your read-only DB. Then it opens a PR with the fix. You wake up to a diagnosis and a draft PR instead of a red channel and a 3am panic about which century it is. And because it's open source and fully customizable (prompts, memory, tools, repo), you're not renting our agent. You're building your own debugging teammate, on your data, on your infra. Cloud option if you'd rather not deal with the deployment. We'd love your feedback, and we're onboarding early teams by hand this week. Ask us anything below.
congrats on the launch 🚀
Maker

 thanks man, appreciate!

Cutting out the need for another dashboard and working directly inside Slack is huge. Receiving a ready PR with a clear diagnosis instead of a late night panic search is a massive relief for dev teams.

Congrats guys, going from a Sentry alert straight to a fix PR is sick, what changed after previous launch?

Maker

 thank you Igor! We just had many discussions with great companies where they weren't as interested with the improved telemetry experience. But they still wanted the issue triage and bugfixes.

So we decided to provide our harness as a separate product with a simpler onboarding and make sure more people can use what we're building at Superlog!

Maker

 main update is that you can now use Superlog Responder on top of your telemetry, while with Superlog you had to install Otel with us! Now it's literally 2 clicks, and you're bug-free!

Sounds like a cool tool! Great that you're offering a free tier for folks to test out. How are you approaching growth? What's your ICP?

Maker

 right now the vast majority of our users are startups, from seed to series B. technically anyone using sentry or datadog, who want to easily take care of fixing whatever breaks!

Maker

Excited to launch Respondeer today! Please give Responder a try - the world needs less bugs and more sleep for SRE engineers!

 looks awesome, we actually built smth similar internally, however, it'd be great to try and see if yours works better

Maker

 awesome! please try it out - many of our users have built a very similar tool internally. Responder is a framework so that you don't have to maintain them. Our users choose Responder because it that knows when to escalate, cuts down noise and provides brief and clean issue messages.

We spend a lot of time tuning how the model behaves, investigates and how it communicates its findings.

Maker

  let's do it!

Love it, been using Responder in beta for few days now and it's awesome, well done team

Maker

 Thank you Camille! Super happy to get good feedback from you :)

Huge! How do I make sure it doesn’t create more bugs?
Maker

 Thank you! We're obsessing over the quality of our triage and fixes. a big thing for us is to make issue messages relevant and grokkable.

for us, it's fundamentally an issue of taste and talking to users!

Maker

main goals are:

  • avoid false positives and unnecessary noise

  • make sure the messages are relevant

What has worked best so far:

  • keep a small set of test cases that you can run right away to see the impact of any prompt or harness changes

  • read the full traces carefully (a tip from Boris Cherny and mentioned by Arseniy in a previous comment) to catch tool or environment bugs before changing the prompt

Our prompt is actually very simple. Agents are already smart. They mostly just need the right tools.

💡 Bright idea
Looks cool! How do you test and iterate on prompts / models? I'm guessing that if the triaged alerts are noisy it's not that much better than usual Sentry/Datadog alerts
Maker

 it's a tough process! our main criteria is:
- preventing false positives and noise
- making sure that the messages are relevant

What works the best so far is:
- having a handful of test cases that you can immediately launch and see the effect of any harness / prompt improvements
- reading full traces thoroughly (Boris Cherny's tip!) to rule out any tool / environment bugs.

our prompt is actually fairly minimal. agents are smart, they just need the right tools

Interesting approach! One question: how do you prevent the AI from proposing fixes that are technically correct but don't align with a team's architecture or coding standards? False positives are pretty common with automated debugging tools, so I'm curious how you handle that

Maker

 Great question Matheus!

  • For coding standards: We allow you to fully customize your agent, including adhering to any architecture / guidelines in the repo.

For false positives: this is our core mission. Responder will only raise an issue if code + telemetry indicates real impact beyond reasonable doubt. I'm personally obsessing over all issues we raise (we instrument our own prod with Responder obviously).

Using an AI SRE like Responder is actually a great way to cut down on noise from alerting systems!

 That makes sense. I like the emphasis on reducing noise instead of just automating debugging, that's usually the harder problem to solve. Thanks for the detailed explanation, and good luck with the launch!

💡 Bright idea

Great product! I was checking out your MCP here and the url tells me that you support on prem or BYOC?

Awesome boys, love the agent approach on top of logs, congrats on the strong +90% PR-merge rate

Maker

 Thank you so much Victor!!

12
Next