An AI-native creative studio. Generate image, video, and audio on an infinite node canvas, design finished artwork in a vector editor, and let AI agents create alongside you — all in one browser workspace.
Prompt-to-media generation, a real design editor, and an API for AI assistants share one account, one asset library, and one credit balance. Create a shot, refine it, compose it into a finished deliverable, and share it — without leaving the tab.
A node graph where each node turns a prompt, references, or upstream nodes into an image, video, or audio clip.
A vector editor for turning generated assets into posters, ads, decks, and diagrams — export-ready.
A self-describing API so an AI assistant can generate media and author fully-editable designs on your behalf.
Every node walks the same short path — decide what you're making, how, with which model, then generate. Nodes wire together, so one output feeds the next.
Image, video, or audio.
General, Edit, Animate, Transition…
Only models valid for that function.
Aspect ratio, resolution, duration.
Result lands on the node.
General (text→media), Edit, Animate (image→motion), Transition & Stitch between frames, and Act & Speak — a character that moves and lip-syncs.
An image node feeds a video node; a character feeds a talking clip. Outputs flow downstream, so a whole campaign lives on one board.
Showcase (a product → a full ad set), Storyboard (a scene per beat), and Carousel (consistent slides) build sequences in one pass.
Guide results with reference images, interpolate between a start and end frame, or use reference-to-video — the controls adapt to each model.
Layer overlays, captions, and transitions over a clip, then render — a lightweight editor built into the node.
Turn a prompt or image into a walkable 3D environment and capture frames from it back onto the canvas.
The model catalogue is configurable — leading image, video, and audio engines side by side — and your recurring people, products, and brands are saved once and reused everywhere.
Generate & edit with reference images, masks, and custom aspect ratios.
Async jobs with live progress — plus talking avatars & lip-sync from an image + audio.
Text-to-speech with a multi-voice screenplay syntax for scripted dialogue.
Clone a person's likeness + voice from a photo and a short clip — reuse them across shots.
Saved product shots and brand profiles — colours, fonts, tone — that steer every generation.
One-click background removal, upscaling, and logo extraction on any asset.
Drop your generated assets onto artboards and build the actual deliverable — social posts, ads, posters, multi-slide decks, and diagrams — then export or share it.
Shapes, text, images, gradients, shadows, and freeform SVG paths — with layers, grouping, snapping and isolate mode.
Multi-slide presentations, flowcharts and org charts with smart connectors that stay attached, and data tables.
Ship to PNG, SVG, or PDF, or present full-screen — pixel-accurate to what's on the artboard.
Public view links, per-slide comments, and role-based collaboration — invite a client or a teammate to edit.
A full REST API mirrors the app, so an agent authenticates with a key and does the work — generating media, wiring up projects, and authoring complete, editable designs. One call to a self-describing guide teaches it the whole system.
The agent guide endpoint returns the whole system — endpoints, models, credits — so an agent learns it in one request.
The agent writes a compact layout spec — like HTML — that compiles into real, editable vector layers: generated imagery, uploads, brand fonts, multi-colour type.
It renders a flat proof server-side to QC, revises in place (no duplicates), and hands back a public share link.
Because the API is the interface, an agent can take a message — say, over WhatsApp — and return a finished, editable design.
Nothing to install — runs in the browser, with a cloud backend and per-user auto-save.
Tiered access — unlimited internal use, or metered by a credit balance for members.
An admin-driven catalogue & pricing — new AI models are added without shipping code.
Prompt it, refine it, design it, and share it — or hand the brief to an agent and get a finished, editable deliverable back.
Open LabX Flow · canvas.labxco.com →