Shuffll
Media & VideoShuffll is an AI video creation platform: give it a topic, web page, Google Doc, or script and it writes the storyline, sources the media, applies your branding, and renders a video. Agents create and export projects, manage brand assets, pick from the template library, and react when a project is ready.
What This Integration Enables
Shuffll is a video factory organized around brand state rather than around clips. Two thirds of its surface has nothing to do with generation: organizations, workspaces, branding, media libraries, folders, and templates. That is the tell. The interesting object here is not the video, it is the workspace, because the workspace is what decides how every video it produces will look and sound. FlowRunner agents create projects from a prompt, a public web page, a Google Doc, or a finished script, follow the generation through its states, and pull the finished MP4 and streaming manifest when the export lands.
The vendor's fastest path is Create Project with Auto Enhance, Auto Export, and a Webhook URL, which runs generation, the branded enhancement pass, and the render in one call and reports back once with the links. That convenience is real, and it also removes every checkpoint between a prompt and a finished branded film. FlowRunner exists to put one of those checkpoints back where it earns its place, so the agent handles the assembly and a person holds the decisions that reach an audience or repaint a library. That is what human-in-the-loop means as a default rather than a speed bump.
Without FlowRunner
With FlowRunner
Use Case Scenarios
A brand refresh that lands deliberately
The brand team publishes a new palette. The agent calls List Organizations to map every workspace, reads each one's current state with Get Branding, and builds a diff. Because Update Branding changes one named property per call and the value's shape depends on which property it is, the agent stages the writes rather than firing them: colors as a color object, logo as a URL string, company as a company object. It posts the diff and the queue of projects due to render this week to Slack, and only after the brand owner names the scope does it run the writes and confirm each one with a read back.
The content calendar renders itself
A row in Notion or Airtable flips to ready. The agent picks up the source, a public article URL or a Google Doc, chooses a storytelling technique and tone from Get Storytelling Techniques and Get Tone Of Voice Options, and calls Create Project. It follows Get Project Creation Status through STRUCTURE and GENERATING_IMAGES, waits on Get Enhancement Status for the branded pass, then calls Export Project. When Get Export Status returns the direct MP4, the agent hands it to the publishing flow and updates the calendar row with the link.
Client footage folded into generated scenes
An agency receives raw footage and product shots from a client. The agent creates a folder with Create Asset Folder, uploads each file from its public URL with Upload Asset, and moves anything that landed in the wrong place with Move Assets. The uploadPath of each asset is what a project's scenes reference, so the next Create Project call can pull real client footage into an AI generated storyline instead of stock. Delivery links go back to the client through Gmail, and the account manager sees the render land in the project channel.
Human-in-Loop Highlight
Update Branding looks like a settings write and behaves like a distribution decision. It accepts either a workspace id or an organization id, and it changes one property per call, so the same operation can retint a single client's videos or every workspace under the organization depending on which id the agent supplies. Nothing in the response distinguishes those two outcomes. The videos already rendered keep the old brand, the videos rendered afterward carry the new one, and any that shipped in between carry whichever half of the change had landed. So the agent never picks the scope. It posts: "Palette change staged. Applying to organization 'Northgate' would repaint 6 client workspaces with 14 projects queued to render this week. Applying to workspace 'Northgate Retail' alone would repaint 3. Which scope, and do the queued projects wait until the logo write lands too?" The brand owner answers, the agent runs the property writes in order and reads each one back. The staging was the agent's work. The blast radius was not its call.
Agent Capabilities
29 actionsProjects
5- Create Project Creates an AI generated video project from a prompt, a public web page, a Google Doc or Slides deck, or a finished script. Creation is asynchronous and returns a project id while the storyline is still being written.
- Get Project Reads a project's name, workspace, and current state as it moves through STRUCTURE, GENERATING_IMAGES, EDITING, and DONE.
- Get Project Creation Status Reports how far the AI has got with a newly created project, which is the check a flow polls before enhancement begins.
- Delete Project Permanently deletes a project and everything generated for it. Export links already handed out stop resolving, so cleanup flows treat this as final.
- List Workspace Projects Lists every project in a workspace with its id, name, status, and creation date. The way to find an existing project before exporting it.
Rendering
3- Get Enhancement Status Reports progress of the automatic enhancement pass that makes a video branded and polished. Shuffll signals "still running" with an HTTP 202 rather than a field, which this operation surfaces as an explicit In Progress flag.
- Export Project Starts rendering a finished project to a video file and returns the ids used to follow it. Enhancement has to have completed first.
- Get Export Status Reports export progress and, once finished, returns the direct MP4 and the MPEG DASH manifest for streaming.
Organizations and Workspaces
4- List Organizations Lists every organization the key can reach, each with its workspaces and branding. The starting point for discovering the ids the rest of the API needs.
- Get Organization Reads one organization in detail, including every workspace with its asset, project, template, and user counts.
- Get Workspace Reads one workspace: its counts, its branding, and its stored video prompts.
- Update Workspace Renames a workspace. Name is the only field this route accepts, so logos, colors, and company details go through Update Branding instead.
Branding
2- Get Branding Reads the branding a workspace or organization applies to its videos: logos, palette, company details, pronunciations, and about text. The read that turns a branding change into a diff.
- Update Branding Updates branding for a workspace or an organization, one named property per call, with the value shape depending on which property is named. The operation this page's human gate exists for.
Assets
6- List Assets Lists a workspace's media library, optionally narrowed to one media type or folder. An asset's uploadPath is what a project's scenes reference.
- Upload Asset Uploads a file into the media library from a public URL. Images, video, audio, and PDFs are accepted, and a DASH streaming version is generated automatically for video.
- Delete Assets Removes one or more assets by id. Permanent, and a project that referenced a deleted file loses that media.
- Create Asset Folder Creates a folder in the media library. The name must be unique within the workspace, and the folder has to exist before assets can move into it.
- Move Assets Moves assets into another folder, which is how an import gets organized after the fact.
- Rename Asset Changes an asset's display name. The stored file and its URL are unchanged.
Templates
3- List Templates Lists the account's templates. A template fixes the scene order, the scene types, and their configuration.
- List Workspace Templates Lists the templates belonging to one organization and workspace rather than the account default set.
- Get Template Reads one template in full, including scene ids and the dynamic fields each scene exposes. The operation to run before writing per scene overrides.
Creative Configuration
6- Get Storytelling Techniques Returns the narrative structures Shuffll can build a script around, each with what it is, why it works, and an example opening line.
- Get Tone Of Voice Options Returns the narration tones a project can be created with, such as Natural, Professional, or Energetic.
- Get Video Tags Returns the video categories. The chosen tag steers the script and the media selection, so it is more than a label.
- Get AI Voice Options Returns the available voiceover voices and their tone styles. Each voice carries a voiceType that has to be sent back unchanged.
- Get AI Avatar Options Returns the AI presenters that can appear on screen, each with a preview video to check before choosing.
- Get Music Library Returns the background music tracks. The value a project needs is the track's relativePath rather than its id.
Triggers
1 triggersEvent Triggers
1- On Project Ready Fires when a project in the workspace reaches DONE, meaning generation and enhancement have finished and it can be exported. The first poll records what is already finished and emits nothing, so an existing library does not flood the flow. For export completion specifically, set a Webhook URL on Create Project and Shuffll calls once with the video links.
Frequently Asked Questions
What can FlowRunner do with Shuffll?
FlowRunner agents can run Create Project, Get Project, and Get Project Creation Status in Shuffll, plus 26 more actions.
Does connecting Shuffll to FlowRunner require OAuth?
No. Shuffll connects to FlowRunner with an API key, no OAuth flow required.
Can Shuffll trigger a FlowRunner workflow automatically?
Yes. Shuffll supports 1 trigger that can start a FlowRunner workflow automatically.
Other Media & Video integrations
Start building with Shuffll
$100 in credits. No card required. Connect in minutes.