Flows
Design and automate complex multi-modal AI pipelines on an infinite visual canvas
Overview
Flows provides a node-based visual workspace for orchestrating end-to-end creative pipelines. While Studio uses a linear timeline for precise editing, Flows allows you to map out production logic on an infinite canvas.
With Flows, you can chain together over 35 image and video models alongside ElevenLabs’ industry-leading speech, music, and sound effects models. This modular approach enables you to iterate on specific pipeline steps without regenerating the entire project.
Key capabilities
- Visual pipeline builder: Map complex multi-step workflows on an infinite canvas
- 35+ generative models: Access image, video, speech, music, and sound effects models in one interface
- Non-destructive iteration: Modify individual nodes without regenerating downstream content
- Modular composition: Layer and combine outputs from different models
- Template library: Start from pre-configured pipelines for common use cases
- Studio integration: Export finished flows directly to Studio for timeline editing
Guide
Add nodes
Right-click anywhere on the canvas or use the toolbar at the bottom to add a node. Nodes represent specific models or utility tools.
Connect your workflow
Click and drag from an output port of one node to the input port of another to chain them together. For example, connect a generated image to a video animation node.
Execute and iterate
Click the Run button on individual nodes to generate content. Re-run a single node to test variations without affecting the rest of the chain.
Export or move to Studio
Once your flow is complete, export the final assets directly or send the project to Studio for fine-tuned timeline editing.
Starting options
Flows is designed for both exploratory creation and structured production pipelines.
Quick start
Start from scratch
Start from scratch
Begin with a blank canvas. This is ideal for building custom pipelines from the ground up or experimenting with complex multi-step generation chains.
Use a template
Use a template
Select from preset shortcuts at the center of the canvas to load pre-configured node structures:
- Perfume Ad: A template optimized for luxury product visuals and voiceovers
- E-commerce Store: Designed for creating rapid product demos and variations
- Yacht Promo Shoot: A multi-modal setup for high-end travel and lifestyle content
- & more: Additional templates are continuously being added
The node system
Nodes are the building blocks of a flow. Each node serves a specific purpose, from inputting text to generating high-fidelity video.
Generation nodes
Generation nodes
These nodes house the generative models that create your assets:
- Text to Speech: Generate narration using our latest models, including v3
- Image Generation: Choose from 35+ industry-leading models to create visuals
- Video Generation: Animate images or generate video clips directly from text
- Music & SFX: Add atmospheric soundtracks or targeted sound effects to your project
See our capabilities overview for detailed information on each model type.
Processing & utility nodes
Processing & utility nodes
Use these nodes to refine and combine your content:
- Lipsync Generation: Chain a speech node with a video node to create perfectly synced talking heads (powered by OmniHuman and Veed)
- Composition: A centralized preview node where you can layer audio and video to view the final output
- Upscale: Increase the resolution and quality of your generated visual assets
- Mix with Video/Audio: Merge separate streams into a single file for final production
Execution and iteration
Unlike linear editors, Flows allows for non-destructive iteration.
Partial generation: If you want to change the voiceover but keep the generated video, you only need to re-run the Text to Speech node. Only the subsequent nodes in that specific path will need to be updated.
Batch variations: Because the structure is saved as a flow, you can quickly swap out a single prompt or character and re-execute the pipeline to generate dozens of creative variations for A/B testing or multi-language campaigns.
Credit usage: Generating content within a node costs credits just like our standalone tools. Re-running a node will trigger a new generation and a new credit charge.
Export and final production
Once you are satisfied with the output of your flow, you have two primary options:
Download assets: Export individual files or the final composition from the result nodes.
Export to Studio: Send your assets directly to Studio if your project requires precise timeline-based adjustments, captions, or manual track layering.
Programmatic execution via API for mass creative production is planned for a future release.
FAQ
What models are available in Flows?
Flows supports over 35 image and video models alongside ElevenLabs’ text to speech, music, and sound effects models. See our models page for a complete list.
How does Flows differ from Studio?
Studio uses a linear timeline for precise frame-by-frame editing with tracks for video, audio, music, and captions. Flows uses a node-based canvas for building multi-step generative pipelines. Use Flows to orchestrate complex generation chains, then export to Studio for final editing.
Can I save and reuse flows?
Yes. Each flow is automatically saved as a project. You can reload a flow, modify prompts or settings, and re-execute to create variations without rebuilding the entire pipeline.
Do I need to regenerate everything if I change one step?
No. Flows supports non-destructive iteration. You can re-run a single node, and only downstream nodes connected to that path will need to be updated. Unconnected branches remain unchanged.
How are credits charged in Flows?
Each node generation costs credits based on the model and settings used, identical to standalone tool pricing. Re-running a node triggers a new generation and a new credit charge. See our pricing page for details.
Can I export flows to other formats?
You can export individual assets or the final composition as downloadable files. You can also send your flow outputs directly to Studio for further editing and timeline-based adjustments.
Is there an API for Flows?
Programmatic execution via API is planned for a future release. This will enable mass creative production use cases, such as connecting a CMS to a flow for automated content generation.
What happens if a node fails?
If a node fails during execution, only that node and its downstream dependencies are affected. You can re-run the failed node without affecting other branches of your flow.
