After a May spent turning the library into a workspace, June was about who gets to do what, and when the work should happen on its own. The theme this month is control: precise access for people, automatic action for agents, and a handful of dials that let each organization tune the AI to its own taste. Here is the tour.
Permission-based roles, defined down to the action
For most of Ceivo's life, access came in a few fixed sizes. June replaced that with permission-based roles: a database-backed system where admins grant specific capabilities rather than handing out a one-size role and hoping it fits. Want a reviewer who can comment and approve but never delete, or a contributor who can upload but not reshare outside the org? Now you can build exactly that. The whole authorization layer was rebuilt underneath, the old catch-all owner role retired, and the interface itself now shows or hides controls based on what the person in front of it is actually allowed to do. Less clutter, fewer accidents, and an access model that maps to how real media teams are structured.
Agents that fire when work happens
Until now, Ceivo's agents ran when you asked them to. June added event-based triggers, so an agent can fire the moment something happens in the workflow: when a file finishes ingesting, when a review starts, when a review is completed, or when a share goes out. Point a brand-standards check at "review start" and it runs before a human ever opens the asset. Hang a notification agent off "share start" and the right people hear about it automatically. This is the step that turns the agent catalog from a set of tools you reach for into a layer that works while you are doing something else, and it is what the spring agent catalog was always building toward.
PDFs join the library
Ceivo has analyzed video, audio, and images for years. As of June it analyzes PDFs too, with page thumbnails and full document analysis, so the scripts, decks, rights documents, and style guides that live alongside your media are searchable and governed in the same place as everything else. The library stops being a video tool with attachments and starts being the system of record for the whole production.
Pull any scene as its own clip
Scene detection has always known where the moments are. June lets you download a single scene as an extracted clip, with optional pre-roll and post-roll padding, straight from the player or the API. Find the beat you want, pull it as a standalone file, and hand it off. No round trip through an editor, no scrubbing a full master to trim one section.
Tuning the AI to your operation
A cluster of June releases hand the AI's behavior back to admins. Versioned analysis prompts give your organization's analysis instructions a full history, one-click rollback, and an audit trail, so you can change how the AI reads your content without losing the version that worked. Scene-detection sensitivity profiles let you pick from predefined presets instead of living with one global setting, which matters when fast-cut promos and long-form interviews need very different eyes. And configurable video thumbnails let you choose how a clip's poster frame is picked: AI-selected, or simply the first non-black scene. Small switches, but together they mean the AI behaves the way your operation wants rather than the way a default decided.
Top 10 notable improvements
Beyond the headline features, ten smaller items worth calling out:
- Multi-season search ranges: search across a span of seasons in a single query, for example seasons 1 through 5 at once.
- Editorial file formats: upload EDL, FCPXML, and OTIO files so edit decisions and timelines live next to the media.
- Org-configurable upload types: admins decide which file extensions are allowed into the library.
- Linkable markers: a marker can now point to a web reference or to another file inside Ceivo, connecting related content directly.
- Permission-driven interface: controls appear or disappear based on the new role system, so people only see what they can use.
- Recursive connector scoping: granting a connector access to a folder now includes its subfolders automatically.
- Leaner agent search: batch search and a compact response mode cut the round-trips and payload size for MCP and API calls.
- Optional emails on share: recipient email addresses are now optional in the sharing flow.
- Curated folder descriptions are safe: the AI no longer overwrites a description a person set, or refiles assets you already placed.
- Diagnostics drift detection: the diagnostics layer now spots document drift and backfills through events to keep analysis current.
Why this month mattered
May made Ceivo a place teams work. June made it a place teams can trust to enforce the rules and to act on its own. Permission-based roles mean access finally matches the shape of a real organization, with Atrium and the broader review workflows resting on top of it. Event-based triggers move agents from on-demand helpers to an always-on layer of the platform. And bringing PDFs in, plus the controls for tuning analysis, push Ceivo further toward being the single governed home for everything a production touches, not just the footage. Read last month's release for where this thread started.
What's next
Want early access to something you saw above, or curious how permission-based roles or event-triggered agents could fit your operation? Reach out and we will take it from there.