Skip to contentMasha Sidorova

Prime Focus

A media library people find their way in, on desktop and on the phone

CLEAR is the platform where studios keep, review and deliver their films, shows and promos. People got lost in it, relied on documentation, and rarely noticed the AI that was already there. We rebuilt the navigation, brought the AI to the front and made the library work on a phone.

Problems

Teams kept a studio's whole library in CLEAR, and still got lost in it.

Interviews with the teams and our audit agreed. People used the tool with the documentation open, terms like Ingest and Work Order confused them, and the executives at the end of the chain barely opened it. One of them said he hadn't watched anything on CLEAR in years. The problems on the user side were:

  1. Navigation that behaved differently from screen to screen. The title and the cover of the same asset did different things, the Action menu was unpredictable, and the Bins panel went unnoticed.
  2. AI features that existed but were never shown.
  3. A cluttered library with hidden filters, no memory of past searches, slow uploads and broken drag-and-drop.
  4. No way to check content away from the desk, and an experience executives found too hard to use.
On this pageSolution

Solution

  1. A new information architecture with a menu that stays in one place, and one rule per element, so the same click does the same thing everywhere.
  2. AI marked with one label and offered where the work happens, from transcripts and metadata to scene segments.
  3. One library flow: search with recent queries and a scope switch, filters with counts, color statuses, drag-and-drop into folders, bulk actions, and the Draft always at hand.
  4. A mobile web app for quick checks on the go, with search, asset details and segments. The scope stayed small on purpose, so it can grow with use.
Summarize with AI

Decision 2

The AI was already there. People just couldn't see it

CLEAR already transcribed, tagged and cut content into segments, but nothing in the interface said so. We gave AI one label and put it where people make decisions: next to the file it processed, on the metadata it filled in, and on the scene segments it found. A person can check what the AI did and correct it in place.

Decision 3

Find it, sort it, put it in the draft

The core job in CLEAR is to find the right material and collect it. Search remembers recent queries and lets people choose between this library and everywhere. Filters show how many files of each type there are before anyone clicks. Files drag straight into a folder, several can be handled at once, and the Draft bar stays at the bottom of every screen, so the story being built is always one click away.

Media library, Lionsgate: the file list with statuses, the search field on top and the Draft bar at the bottom
Search opens with recent searches and a scope switch: Media library or Everywhere
27 results. The filter opens by type, with a count next to every type
A file is dragged from the list toward the folder tree
The file drops into the Idents folder, the target is highlighted
4 files selected: the actions bar replaces the toolbar, with Cancel at the end
The same library as a grid: previews, durations, sizes and the AI label on processed files

Decision 4

A phone for quick checks, not for the whole job

CLEAR had no mobile version. Research pointed to a narrow set of tasks people do away from the desk: find an asset, check its details, look through its scenes. We built only that, with search and a scope switch on top, and kept the rest for the desktop. Because the mobile app is new, its scope can grow with what people actually use.

Search on the phone: typing big shows matching titles in the media library
The scope switch: search everywhere or only in the media library
Media library, Lionsgate: files with dates, statuses and the AI label; Export, Add folder and Upload on top, the Draft at the bottom
Asset view, Silicon Valley: the player, then details with the AI-generated general information
Segments: the scenes of the episode with time codes and AI summaries, ready to pick

See other cases