Launch →

Build visual
AI workflows for
scalable content.

Mosa is the visual AI workflow platform built by Ambassadors, the studio behind Cube and Vocoda, for creative and marketing teams producing content at scale.

Mosa launches on 1 October 2026.

Why Mosa

We didn't set out to build another AI tool.

Mosa started inside Ambassadors' own production work, when scattered prompts and disconnected tools couldn't keep up with real briefs, real markets, and real deadlines. So we built the system we needed, and kept building it until it was ready for other teams too.

Bring every piece of your content together

Scattered prompts and disconnected AI tools slow you down. Mosa lets you map out your entire creative process visually, connecting different AI models, formats, and channels into one cohesive workflow. Take full control over how your brand's story is generated and distributed at scale.

A Mosa workflow on the canvas: a Gemini quality-check node reviews a generated image, and a conditional routes it on only if it passes.
01

Scale output, not complexity

Run global campaigns across markets, formats, and audiences simultaneously, and if anything interrupts, every workflow picks up exactly where it left off.

02

Visual logic without code

A canvas-first editor with live collaboration. Every branch, loop, and condition is something you can see and touch.

03

All models and formats on one canvas

Every format your campaign needs, text, image, video, and audio, in one place. Work with any model or provider in a single, fully traceable workflow, so your team stays in control from idea to output.

01

Intelligent workflow logic

Set your own rules for quality checks, conditional routing, and automated retries. You dictate exactly how your content is generated, ensuring every single output meets your strict brand standards before moving forward.

02

Predictable & transparent costs

No hidden fees. Track exactly what each workflow costs in real-time. With itemized billing, you always know exactly where your AI production budget is going, keeping your campaigns profitable.

03

An agent that builds with you

Describe what you need and the Workflow Agent builds it on the canvas, live, as a collaborator. It tunes nodes, debugs failed runs in plain language, and learns the processes you teach it, all within your budget.

04

Every asset, fully traceable

Ask any file who made it, when, and what produced it. Step back through what earlier runs of a node generated, so approvals and audits never depend on guesswork.

Integrations

The best models, in one workflow.

Mix providers freely in a single workflow: write with Claude, generate images with GPT-Image, animate with Veo, and voice it with ElevenLabs. When a new model comes out, switching to it is one dropdown away.

AI providers

12

  • OpenAI

    GPT · GPT-Image · Whisper

  • Anthropic

    Claude Opus · Sonnet · Haiku

  • Google

    Gemini · Nano Banana · Veo

  • xAI

    Grok

  • BytePlus

    Seed · Seedream · Seedance

  • Kling

    Image & video generation

  • Luma

    uni-1 · Ray

  • Fal.ai

    Flux · Recraft · Ideogram · Topaz

  • ElevenLabs

    Voice · Speech-to-speech · Scribe

  • Vocoda

    Voice synthesis

  • Magnific

    Image & video upscaling

  • Serper

    Google Search & Images

Platform integrations

04

  • Google Drive

    Pick the source files a workflow starts from, and deliver finished assets straight into shared folders.

  • Google Sheets

    Drive a whole campaign from a spreadsheet: read briefs and copy row by row, and write results and links back.

  • Figma

    Pull frames, exports, and design tokens from your Figma files so every generation stays on brand.

  • Cube

    Read assets from your Cube library and publish approved outputs back to it, ready for distribution.

Nodes

Building blocks for any production.

Every workflow is made of nodes you drag onto the canvas and wire together. Here are some of the ones teams use most. There are many more for files, arrays, and data shaping.

AI

Generate

Text, image, video, and audio from the models you choose, side by side on one canvas.

  • LLM

    Write copy, turn a brief into prompts, extract structured data, or review outputs against your brand rules.

  • Image Generation

    Create and edit images from a prompt and reference images, with GPT-Image, Nano Banana, Seedream, Flux, and more.

  • Video Generation

    Turn text or a still image into video with Veo, Seedance, Kling, and Luma Ray.

  • Audio Generation

    Produce voice-overs in any language, convert a recording to another voice, or add sound effects to a video.

  • Lipsync

    Re-render a speaker's mouth to match a new audio track, for localized, script-driven talking videos.

  • Upscale Media

    Upscale and enhance images and video to delivery resolution with Magnific, Topaz, and others.

Data

Connect

Bring in briefs, designs, and source assets, and deliver results where your team works.

  • Google Sheets

    Read campaign rows from a spreadsheet to drive a workflow, and write results back to the right row.

  • Google Drive

    Use Drive files as inputs and upload finished assets to the folder you choose.

  • Read Figma

    Import frames as pixel-exact exports, along with the colors and type styles they use.

  • Cube

    Read folders and assets from your Cube library, and write approved outputs back.

  • Web Screenshot

    Capture a live web page as an image, a faithful reference for on-brand generations.

  • String Format

    Build prompts and file names from templates, filling in values from anywhere in the workflow.

Control

Decide

Loops, branches, and reusable building blocks that turn one asset into a whole campaign.

  • Loop

    Run the same steps for every market, format, or product in a list, and collect all the results.

  • Conditional

    Send work down different branches based on a rule, such as retrying an output that fails a quality check.

  • Switch

    Pick one of several inputs by index, such as the right template for each channel.

  • Gate

    Only let data continue when a condition is met, so nothing moves on until it should.

  • Group

    Fold a set of nodes into one tidy block with its own inputs and outputs. Double-click to open it.

  • Workflow

    Embed one workflow inside another. Build a step once and reuse it across your team's playbooks.

Utility

Finish

Deterministic media tools that take generated assets the last mile to delivery.

  • FFmpeg

    Trim, join, overlay, normalize loudness, compress, and convert, from presets or any custom FFmpeg command.

  • HTML Render

    Render HTML and CSS scenes, animations included, to an image or video, frame by frame and reproducibly.

  • Text Render

    Set headlines and captions as crisp images in the font and weight you choose, ready to composite.

  • Resize Media

    Scale images and video to the exact resolution each channel needs, keeping the aspect ratio.

  • Image Crop

    Crop one image or a whole batch to any aspect ratio, from 9:16 stories to 16:9 banners.

  • FFprobe

    Read a file's duration, size, and codecs, and use those values to steer the rest of the workflow.

Come and build with us.

From 1 October 2026 anyone can sign up and get going on their own. Until then we are onboarding teams one at a time. Get in touch and we will set you up with a workspace and credit to try it on real work.