A Backend Your Whole Team Can Actually Use

Directus has always been the collaborative backend. The API flexibility, the database-first architecture, the granular access control. Developers picked Directus because it let them build on their terms without giving up control.

Collaboration requires more than one side of the team showing up, though. If we're being honest, the Studio has been a developer's tool that content teams could use, not a workspace they'd choose.

Editorial workflows required workarounds. Translations required careful schema work. The experience was powerful but it was built for the person who configured it, not the person who uses it every day.

That changes now. With our latest major release (v12), the collaborative backend works for the people who actually manage content, not just the people who set it up.

What's New

Directus now has native editorial workflows, simplified multilingual content operations, a redesigned Studio for non-technical users, and proper authentication for AI-connected workflows. This release also includes a licensing transition that reflects where Directus is headed as a platform.

Every major change maps back to the same question: does this make Directus work for the whole team, or just the developer?

Native Draft and Publishing

Content teams have been asking for a real editorial workflow inside Directus for years. The workaround was always some version of a custom status field: draft, in review, published, managed through a flow or just team discipline. It worked, but it was never native.

That's over. Items now have explicit draft and published states. Editors work on drafts by default, and publishing is an intentional action, not a status field toggle. Published content can be unpublished or archived through a dedicated flow rather than manually flipping a value. Version indicators show up across the Studio (on item pages and in collection views), so there's never ambiguity about which version you're looking at or editing.

New items open in draft mode automatically. That one detail changes the entire feel of content editing in Directus. You're no longer editing a live record and hoping nothing breaks. You're working on a draft, on your own terms, and publishing when it's ready.

This also lays the groundwork for what comes next. Scheduled releases and approval workflows build on top of this foundation.

Studio Redesign

The Studio updates aren't cosmetic. The navigation, headers, sidebars, and action patterns have all been reworked to support the new editorial workflows and to make the day-to-day experience clearer for non-technical users.

This is the kind of update that's hard to describe in a changelog but obvious the first time you use it. Draft indicators, version selectors, and publish actions are now core UI elements, not buried settings. The sidebar surfaces content state at a glance. Navigation reflects where you are in the editorial workflow. The whole thing feels like a Studio built around content operations, because that's what it is now.

For developers, the Studio improvements translate directly into fewer support requests from your team. When the interface communicates state clearly, people don't need to ask "is this live?" or "which version am I looking at?"

AI Translations

Multilingual content in Directus has always been possible, but the setup required deliberate schema work and a developer who understood the relational model well enough to configure it properly. That's a real barrier for teams that need to move fast on localization.

Directus now simplifies how translations are configured on collections. The implementation friction drops significantly, which means developers spend less time on setup and content teams can start working in multiple languages sooner.

On top of that simpler foundation, AI-assisted translations let content teams scale across languages faster without routing every string through a manual translation process or a third-party TMS.

It's not a replacement for professional translation on high-stakes content, but for the volume of content that just needs to be accurate and available, it removes a bottleneck that slows down nearly every multilingual project.

JSON Filtering

This one is for the developers. JSON fields in Directus have always returned the entire blob, which meant developers had to pull everything to the client and parse out what they actually needed. There was no way to query or filter on a specific value within a JSON structure through the API. You either worked around it with custom endpoints, raw SQL, or schema redesigns that split the JSON into relational fields.

Directus now supports dot-notation path queries on JSON fields, with filtering at the database level. You can reference a specific nested value and filter against it directly. No client-side parsing overhead, no workarounds.

It's the kind of improvement that doesn't make headlines but changes how developers feel about building on the platform. When you store structured metadata, form submissions, feature flags, or configuration in JSON fields (which plenty of real-world projects do), you can now query that data the way you'd expect.

MCP OAuth 2.1

Directus already supports MCP (Model Context Protocol) for connecting AI agents and tools to your data. But without OAuth, every MCP client needed a static API token. That means no user identity, no consent flow, no way for third-party AI tools to connect on behalf of actual users with actual permissions.

MCP now supports OAuth 2.1. AI agents authenticate as real users with the same access policies that apply to everyone else in the system. That's a meaningful jump. It's the difference between "we demoed an AI integration" and "our security team approved an AI integration for production."

For teams building MCP-connected workflows, this removes the biggest blocker to enterprise adoption: the security review. Dynamic client registration, proper consent flows, and user-scoped permissions mean AI workflows get the same governance as everything else in your stack.

Licensing Transition

This release introduces a new licensing model, transitioning from BSL to MSCL (Monospace Community License). This includes technical license key enforcement and represents a breaking change for existing instances.

Directus is becoming a platform that teams depend on for production content operations. The license change reflects that. It keeps the platform accessible to builders while creating a clear value exchange for organizations that get real operational value from Directus.

The license introduces seat caps and collection limits that scale with your plan tier. Core remains free with reasonable limits (3 seats, 50 collections). Team and Enterprise tiers expand those limits with add-on flexibility. Collections can be deactivated rather than deleted, which matters for teams that need to adjust their usage.

For existing customers upgrading: You can upgrade without a license key. If you do, the instance enters a 30-day grace period that prompts you to contact support at licensing@directus.io for your key. We'd recommend reaching out before upgrading so you can set the key as an environment variable and skip the grace period entirely. Keys can also be entered through the new Settings > License page in the Studio.

Looking Forward

Scheduled releases and approval workflows are next, and they're only possible because this release builds the editorial infrastructure underneath. The Studio will keep evolving around these workflows. The AI story gets stronger as MCP adoption grows.

If you've been building on Directus and wishing your content team could do more without filing tickets, this is the upgrade that makes that real. If you've been evaluating Directus and wondering whether it's ready for your whole team, now's the time to look.