← All posts

June 21, 2026

Self-hosted vs hosted agent observability — what you actually pay for

“Self-hosted” sounds like the cheap option, and on day one it is. You run one command, point your agents at it, and you’re live in ten minutes. The price tag is zero. So why would anyone pay for a hosted version of the same open-source software?

Because the bill for self-hosting isn’t the deploy. It’s every day after.

What deploy actually costs

Standing up PulseDeck is genuinely easy: a container, a Postgres database, a reverse proxy. The work that doesn’t show up in the quickstart is the work that never ends:

None of this is hard. All of it is continuous, and it competes with the work you’re actually paid for.

When self-hosting is the right call

Plenty of teams should self-host, and PulseDeck is built so they can — every feature, no paywall, your data in your own Postgres. If you already run infrastructure, have an on-call rotation, and want the data inside your perimeter for compliance, self-hosting is the obvious answer. The AGPL license means you can fork it, audit it, and own it.

When you’re really buying time

For a small agency or a lean team, the math flips. The monthly fee for a hosted plan is almost always cheaper than the engineering hours you’d spend operating the same thing — and far cheaper than the incident you didn’t staff for. You’re not paying for features. You’re paying for not having to operate it.

That’s the whole tradeoff. The software is identical; the difference is who gets paged.


Either way, start the same: self-host free in one command, or read the Quickstart to see it running with live data.