I Built Ravn Because Sentry's Pricing Made Me Mad

Half a year ago, I started building a side project. Nothing super special, a simple Flask API with a database and a frontend.

Early on, I wanted to set up error monitoring, so I signed up for Sentry.

The free tier was pretty generous. 5000 events a month. But my project became bigger than expected, so I used it all up in 3 days.

The next pricing tier was $26 a month. For fifty thousand events.

Fifty thousand events for a side project where I'm not making money yet.

I was annoyed, because I was looking at an error monitoring service costing more per month than my hosting.

So I built something simpler.

What Ravn actually is

It's error monitoring for Python. When your code breaks, Ravn captures the exception, groups it with similar errors, and shows you what happened. Pretty simple actually.

I also added AI root cause analysis, basically due to peer pressure, because everyone is doing it right now. Now, when an error happens, it tries to tell you why, not just that it happened.

The setup is two lines of code:

import ravn

ravn.init(api_key="your_api_key")

That's it, every unhandled exception gets captured. No configuration necessary.

How this came together

I built the backend in FastAPI because async felt right for it. PostgreSQL for storage, pgvector for the embeddings, Redis for caching. React on the frontend because it's what I use all the time and it works.

The Python SDK is on PyPI. Open source. People can contribute if they want.

I spent way too long on the dashboard because I'm very picky about design and I don't want it to look like average vibe code slop.

What I've learned in 6 months

Why people don't switch from Sentry

1. Their whole team ist already there

2. They pay for features they don't use but feel safer having

3. New tools feel risky without special proof

What actually gets people to try Ravn

1. The free tier ist REAL (not a demo)

2. Open source SDK

3. Honest about what's early

The marketing problem

PH is brutal without an existing audience. Reddit/Discord/IH conversations convert better than algorithmically sorted launches.

What still sucks

It's early. The AI analysis works but it's not a new revelation. Sometimes it just tells you the obvious thing. The alerting is basics, but it works. Slack and Discord only right now.

I built this because I was frustrated, not because I'm trying to kill Sentry (that'd be pretty funny actually). Jokes aside, Sentry is great. Extremely solid tool. Just overpriced and too heavy for me.

If you want to build something small and don't want to think about error monitoring, Ravn is there. If you need enterprise features and have a team, Sentry is probably the right call.

The real question

Is there actually a market for Sentry without the overhead? Or are people happy enough with Sentry that they won't switch even if it's cheaper?

I'm genuinely not sure yet.

If you've built a tool and struggled to get initial traction, I'd love to hear what worked for you.