Pre-release products are usually described in the present tense regardless of whether they exist. This post is the correction: a plain account of what runs today, what is being built, and what is still a line on the roadmap.
If you are evaluating Birkly, this is the page to read before the product page.
Shipping now — alpha
The core content model. Collections, entries, fields, sub-collections. Stable, and unlikely to change shape before 1.0.
Templating and rendering. Both connection modes work: the client script on an existing site, and server-side rendering from the project/ folder. This marketing site runs on the second one, which is the most useful thing we can say about its readiness — every page you can click on is a Birkly template resolving against live collections.
Admin, users and roles. Permissions are real and enforced, not a placeholder.
MCP. Read tools across collections, entries, project files and settings. Write tools with approval gating. This blog post was drafted through MCP into the approval queue, which is a slightly self-referential test but a genuine one.
Agent Guide. Products, claims and evidence, queryable by connected agents. The hub view and full WebMCP surface are landing now.
Commerce. Hub publishing, purchases and licence keys run through it — we use it for our own Hub, which is the only reason we would describe it as available rather than beta.
Automation. Workflows with triggers, conditions and actions. Free.
In progress — beta
Project Git sync. Committing the project folder from the admin. Partially working, not something we would ask you to rely on yet.
Developer mode. Deeper inspection of what the template engine is doing when it does not do what you expected.
Insights and analytics presets. Plugin activity and usage reporting.
Site packs and Hub polish. Curated starters with a content model and a design baseline you can own and extend.
Donations via Stripe Checkout. On the sponsors page, wired through Commerce.
Not built
The candid section, because this is the one that determines whether Birkly fits what you need this quarter.
Multi-language. There is no i18n in the core today. We had a blog post up describing side-by-side translations as if they existed; it was written before the feature was, and we took it down rather than leave it standing. When translation lands, we will write about it then.
WordPress import. On the roadmap, not built. Bringing an existing content base across means writing an import against the API today.
Version history. Entries carry a version number; there is no interface for browsing or restoring previous versions yet.
Scheduled publishing, webhooks, offline sync, AI alt text, spam protection. All on the public roadmap, all unbuilt. The roadmap lets you vote on the order, and votes genuinely move things — though hard architectural dependencies keep some items where they are regardless of how popular they get.
Events and Studio plugins. Announced, in development, not available.
What we got wrong since the last update
Two things worth recording.
The first is the multi-language post mentioned above — a feature described in the present tense before it was real. That is the exact failure mode this build log exists to prevent, and it happened on our own blog, which is instructive.
The second is that we underestimated how much of the value sits in reading rather than writing. We built the MCP write path first, on the assumption that AI-assisted authoring was the headline. In practice the interesting behaviour has been agents reading a Birkly instance and answering accurately from it. That reprioritised the Agent Guide considerably.
What is next
Closing out the beta list above, then a 1.0 that is free and open source. No date, because we have not earned the credibility to give one and a missed date is worse than no date.
The roadmap has the full list and the voting. Alpha invites go to the waitlist first. If something on the not-built list is a blocker for you, say so on the roadmap — that is what the votes are for.