1 month after open-sourcing my social media tool: 1.7k stars, and why the hosted version is now $0 for everyone.
Three things converged this month: A Reddit post that went further than I expected, a GitHub repo that grew faster than I planned for, and a decision about pricing that I want to think through publicly.
The context
My girlfriend's agency was paying $400/month for Buffer to schedule posts across ~6 client accounts. As a developer I kept looking at scheduling tools thinking, "this is a CRUD app with OAuth integrations and a cron job. There's no reason it should cost $400/month." So I built one. AGPL-3.0, Django + HTMX + Alpine + Tailwind + Postgres, 12 platform integrations.
I posted placed on post on Reddit and one on HackerNews to get feedback. The result the GitHub repo has crossed 1,700 stars and 350 forks since.
Why the hosted version is now $0 for anyone
Today I'm opening the hosted version of BrightBean Studio to anyone who wants to use it instead of self-hosting.
The reasoning is uncomplicated. We already host the tool internally for my girlfriend's agency for ~$35/month. That bill exists whether one workspace runs on it or 200. The marginal cost of additional small workspaces on the same server is essentially zero, up to a limit we'll watch.
The repo stays the source of truth. AGPL-3.0 license stays. Same Docker image we run is what you'd deploy.
If self-hosting felt like too much friction: Now you don't have to.
I'm aware the $0 model has potential for failure. More on that in the open questions below.
Three things I underestimated about the month after launch
Polish-to-features ratio in real usage
Going in I assumed the feedback loop would be "users want feature X, I ship feature X." That isn't what happened. The bulk of shipped work in the last month has been polish on existing features, not new ones.
A representative pull request: 30+ small changes across the calendar, composer, media library, notifications, and approvals apps. Hex color validation on workspace branding. Permission gating on calendar event endpoints. Workspace scoping on comments, tag input normalization, etc.
None of this shows up on a feature list. All of it is the difference between "works for one user" and "works for many users on different deployments."
The mental shift: When you ship to one user, you build features. When you ship to 1,700 potential users, you finish features.
Self-hosters are the best QA I've ever had
I knew open-source meant self-hosters. I did not know self-hosters would be the best QA team I have ever worked with.
People running their own instance read the code. They file reproductions. They catch bugs that would have shipped without me knowing.
Each issue takes 2-3x longer to answer than a typical SaaS ticket because the bug reports are detailed enough that I have to actually trace the code path. That isn't a complaint. The signal quality is higher than any paid product I have ever used as a customer.
What I would want feedback on
Two open questions for this community:
-
The $0 model. For anyone who has run a similar setup: Where does this break?
-
Support load. The 1.7k stars and 350 forks have come with a sustained increase in issues and PRs. So far I am keeping up. The number I cannot see yet is where this breaks and starts cannibalizing product work. Anyone a year or two ahead on this kind of project, what was your tipping point?
This is a strong open-source story because the traction is not just vanity stars. Self-hosters are forcing the product to mature faster, and the hosted version removes the biggest adoption blocker for agencies and small teams that like the idea but do not want to touch Docker.
The bigger positioning question is whether this stays “open-source Buffer alternative” or becomes lightweight social ops infrastructure for small agencies, creators, and teams. The second frame is much stronger because it moves the product away from being compared feature-by-feature with Buffer and closer to owning the workflow around scheduling, approvals, media, comments, branding, and multi-client publishing.
One thing I’d pressure-test is the name before the hosted version gets more public usage. BrightBean Studio is friendly, but it may feel a bit small for the serious agency workflow you are building around it. If this expands from an open-source scheduler into a hosted social operations layer, the brand may need to feel cleaner and more platform-like.
Xevoa .com would fit that broader direction better because it gives the same product a stronger workflow/platform shell without changing the open-source roots. The product already has real pull. The next question is whether the brand makes it feel like a serious system agencies can build around, not just a generous side project.