Why Self-Hosted First
The bet against SaaS-only
Every observability tool wants your data.
Send your logs to our servers. Upload your errors to our cloud. Trust us with your metrics.
I'm going the opposite direction.
The conversation that changed my mind
I was talking to a CTO friend about observability tools. He was frustrated.
"We can't use Datadog anymore. Compliance won't allow production data leaving our infrastructure."
"What about Sentry?"
"Same problem. We're stuck with self-hosted ELK, and it's a nightmare to maintain."
Then he said something that stuck with me: "I'd pay good money for something that works like Datadog but runs on my servers."
That's when it clicked.
The AI angle makes it even more important
Think about what Brainz Lab does: it gives AI access to your observability data.
Claude searches your logs. Analyzes your errors. Checks your metrics.
If that data is on some third-party cloud, Claude has to make API calls over the internet. Every query is a round trip. Rate limits apply. Network failures happen.
If that data is local, Claude connects directly. Instant responses. No limits. Always available.
Self-hosted isn't a limitation—it's what makes AI integration actually work.
The trust problem
Developers have trust issues now. And they should.
Every week there's a new breach. A new "we accidentally exposed customer data." A new "oops, our employee had access to everything."
Your production logs contain secrets. User IDs. IP addresses. Request payloads. Error messages with stack traces. This is sensitive stuff.
The more tools you send data to, the more attack surface you have. Self-hosted means your data never leaves your control.
I'm not saying cloud tools are evil. I'm saying some teams can't use them, even if they wanted to.
The business model still works
"But how do you make money if people self-host?"
Same way GitLab does. Same way Sentry does. Same way PostHog does.
Self-hosted is free. Download it, run it, use it forever.
Cloud is convenient. Don't want to manage infrastructure? We'll do it for you. That's the paid tier.
Enterprise features cost extra. SSO, advanced analytics, dedicated support, SLAs—these justify enterprise contracts.
The math works. Self-hosted users become advocates. Some convert to cloud when they scale. Some buy enterprise features. The product spreads because it's accessible.
The competitive moat
Here's what's sneaky about self-hosted first: it's really hard for cloud-native competitors to copy.
Datadog's entire infrastructure assumes your data goes to their servers. They can't easily offer self-hosted. Their architecture doesn't support it.
By starting self-hosted, I'm going where they can't follow.
If Brainz Lab gets traction, Datadog can't just flip a switch and offer the same thing. They'd have to rebuild from scratch. That's a multi-year project.
What I'm betting on
I'm betting that the market is shifting.
That developers want control back. That compliance requirements are getting stricter. That the next generation of tools will be self-hosted by default.
Maybe I'm wrong. Maybe everyone's happy sending data to third parties forever.
But if I'm right, I'm early to a very big wave.
— Andres