Frame.io

Connect AI agents to a Frame.io video review workspace. Agents navigate teams, projects, and asset hierarchies, create folders and file placeholders, and read or post review comments, including notes pinned to a specific timestamp in a video.

9 actions API key available

Frame.io website ↗ Platform Documentation ↗ Capability data verified 2026-07-28

A render pipeline finishes a new cut and drops the file with a project reference

Get Me and List Teams resolve the account, then List Projects finds the campaign's project and its root asset

Create Asset builds the dated review folder and registers the file placeholder, returning signed upload URLs

The binary uploads, and the asset ID is logged so downstream systems can find the cut

On a schedule, List Comments sweeps the asset for new review notes and their timestamps

New feedback posts to the production channel as a digest, each note linked with its timecode

The producer reads the feedback digest, decides which notes become edit tasks, and signs off on the reply that goes back to the review thread

What This Integration Enables

Video review generates the most perishable feedback in any creative pipeline: a note pinned to second 47 of version 4 matters intensely for about two days, then either becomes an edit or a miss. Frame.io holds those notes; this connector makes sure they move. FlowRunner agents navigate the account, team, project, and asset hierarchy, create folders and file placeholders so structure exists before uploads happen, and read and post comments, including notes pinned to an exact timestamp in a video. The stance: review platforms should feed production systems, not compete with them. The comment thread is where feedback is born, not where it should retire. - Walk the full hierarchy from account to team to project to asset programmatically - Register placeholders and receive signed upload URLs for render pipelines - Sweep timestamped review notes into digests, reports, and task systems - Post notes back onto assets, pinned to the exact second they reference - Read asset metadata, type, size, source URL, for content-processing flows

Without FlowRunner

Feedback trapped in threads Review notes pile up inside Frame.io while the task board and the edit schedule learn about them days later

Folder structure by willpower Every editor organizes uploads differently, and finding the approved cut of version three is a scavenger hunt

Status meetings as status source Whether the client reviewed the cut is a question someone has to ask in standup

With FlowRunner

Feedback becomes work items Timestamped notes sweep into digests and task systems automatically, with timecodes attached

Structure created by workflow Agents create the same dated folder pattern for every deliverable, and placeholders register before binaries arrive

Review state is queryable Comment counts and completion states per asset feed a live production dashboard nobody has to update

Use Case Scenarios

The review digest that reaches the edit bay by morning

Overnight, the agent runs List Assets across active projects, calls List Comments on every file touched in the last day, and separates new notes from completed ones. Each project's digest posts to [Slack](/integrations/slack) with the note text, author, and timecode, and notes matching action patterns become tasks in [Asana](/integrations/asana) tagged to the editor who owns the cut. Editors start the day with a worklist built from last night's client feedback instead of scrolling threads to reconstruct it.

Render-to-review without a hand on the keyboard

A finished render lands in the pipeline. The agent resolves the project through List Projects, creates the dated review folder with Create Asset, registers the file placeholder with its size and MIME type, and pushes the binary to the returned signed upload URLs. The asset ID and upload record log to [Google Sheets](/integrations/google-sheets) for the delivery team. When the team also publishes via [Vimeo](/integrations/vimeo), a matching folder keeps review and delivery assets aligned. Every deliverable lands in the same structure because no person ever places it.

QC findings pinned where the editor will see them

An automated QC pass runs against a cut's source file, pulled via the download URL from Get Asset, and finds an audio dropout at 2:13 and a black frame at 0:41. Rather than filing a report nobody opens, the agent drafts timestamped notes for each finding. The producer reviews the findings, discards the false positive, and approves the rest, and Create Comment pins each one to its exact second in the review thread, where the editor already works. The QC report and the edit conversation become the same artifact.

Human-in-Loop Highlight

Create Comment publishes into the one thread everyone on a production reads, including the client. A review note is not a log line: it lands with the weight of a creative judgment, and an automated note that is wrong, a false-positive QC flag, a mistimed reference, a tone-deaf phrasing on a director's cut, is read by the people whose trust the review process runs on, and being read cannot be undone. So FlowRunner flows split the comment lifecycle: agents draft notes freely, from QC findings, checklist sweeps, or cross-system status, but a producer approves the batch through a [human-in-the-loop](/concepts/human-in-the-loop) step before anything posts to an asset a client can see. Internal-only assets can run looser by policy; client-facing threads never do. Reading comments, building structure, and registering uploads run at machine speed. Speaking in the review thread waits for the person whose name the production answers to.

Agent processes routinely

Detects exception requiring judgment

Clear match Continues automatically

Ambiguous Routes to human via preferred channel

Human decides

Agent resumes with decision

Agent Capabilities

9 actions

Account

Get Me Returns the profile of the account holder for the developer token, including user ID, name, email, and account ID. The entry point: the account ID feeds List Teams.

Teams and Projects

List Teams Lists all teams within a Frame.io account. Teams are the top-level containers that hold projects; each team's ID feeds List Projects.
List Projects Lists all projects within a team. Each project carries a root_asset_id, the top folder for its files, which is the starting point for browsing with List Assets.
Get Project Retrieves a single project's details, including name, team association, and root_asset_id. Used to anchor asset creation at the project's top level.

Assets

List Assets Lists the child assets, files, folders, and version stacks, directly inside a folder asset, with pagination. Pass a root_asset_id for the top level or any folder ID to browse deeper.
Get Asset Retrieves a single asset by ID with metadata including name, type, file type, size, parent and project IDs, and for files the source download URL and thumbnails. The call QC and processing flows start from.
Create Asset Creates a new asset under a parent folder: type folder for structure, type file to register a placeholder with name, filesize, and MIME type. File placeholders return signed upload URLs for the binary.

Comments

List Comments Lists all comments on an asset, plain-text and timestamped review notes alike, with text, author, timestamp in seconds for timed notes, completion state, and creation time. The feed behind digests and task creation.
Create Comment Adds a comment to an asset, general when the timestamp is omitted, pinned to a specific second of a video or audio asset when provided. The client-visible write this page's human gate reviews.

Frequently Asked Questions

What can FlowRunner do with Frame.io?

FlowRunner agents can run Get Me, List Teams, and List Projects in Frame.io, plus 6 more actions.

Does connecting Frame.io to FlowRunner require OAuth?

No. Frame.io connects to FlowRunner with an API key, no OAuth flow required.

Can Frame.io trigger a FlowRunner workflow automatically?

Frame.io doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.

Start building with Frame.io

$100 in credits. No card required. Connect in minutes.

Start Building Free Talk to us