AI content studio

Build AI content,
block by block.

metagraphy is a content studio you build out of play blocks. Stack text and creative blocks on one canvas, connect them so one block's output flows into the next, and make copy, images, and video — all in one place.

Request an inviteSee how it works

Invitation-only for now — request an invite and we'll bring you in.

Text generationImageVideoVoice ChatWeb searchKnowledge storeConnect blocksFormatsProjects
Building blocks

It all starts with one block.

A block is one instruction paired with its result. That's the whole idea. There are only two kinds to learn — and everything you make is built from them.

  • One block a prompt and the output it produces.
  • Stack them vertically into a project.
  • Connect them so one block's output feeds the next.
one block

Prompt:

Give me one short, memorable launch tagline for our spring single-origin coffee.

Output:

Mornings, reimagined.

Two kinds of block

Words or visuals. Same canvas.

Only two pieces to learn. Write with the latest language models, or generate images and video — then snap them together.

Text block

Copy, scripts and ideas in your format.

#1
Launch caption
gpt-4.1

Prompt:

Write a warm 3-line Instagram caption for our new Ethiopia Yirgacheffe single-origin — floral, with jasmine and citrus notes. Friendly tone, soft CTA, 3 hashtags.

Output:

SNS / InstagramAuto
Spring's first pour is here ☕️ Ethiopia Yirgacheffe — jasmine on the nose, bright citrus on the finish. Come meet your new morning. #AoyamaCoffee #SingleOrigin #Yirgacheffe
  • Voice Chat intake — just talk, it writes
  • GPT-5 family · gpt-4.1 default
  • Output formats + optional web search
  • Grounded in your knowledge store

Creative block

Image & video via cloud generation.

#2
Product hero
Flux.2

est. credit: 1,240

PositiveNegativeAspectFile InputSettings
A kraft bag of single-origin coffee on a sunlit café table, steam rising from a cup beside it, warm morning light, shallow depth of field, editorial product photography.
1:14:59:1616:9

Output

Product hero — generated image 1Product hero — generated image 2Product hero — generated image 3
  • Image and video, all in-block
  • Aspect ratio · negative prompt · reference
  • A curated ComfyUI Cloud model set
  • Output kept in private storage
Text models:gpt-4.1defaultgpt-5.5gpt-5.2gpt-5.1gpt-5gpt-5-minigpt-4o
in progressVoice Chat

What's the story behind this blend — what made you choose this Ethiopia Yirgacheffe?

We tasted it at origin last spring. The jasmine aroma hit first, then this bright citrus finish — we knew it had to be our spring release…

Hold to talk
WebSpeechWhisper APITextChat
Voice Chat

Don't know what to write? Just talk.

An AI interviewer chats with you by voice — it asks the right questions, listens, and turns the whole conversation into ready source text inside your text block. Great for capturing your authentic voice, thinking out loud, or starting from nothing.

  1. 1

    Press talk

    Tap and start speaking. No empty prompt box.

  2. 2

    It interviews you

    An AI interviewer asks the right questions and listens.

  3. 3

    It becomes text

    Your conversation lands as ready source text in the block.

Speech-to-text is built in.

NodeConnect

Place. Connect. Run.

Wire one block's output into the next. A Text block's copy flows down into a Creative block; that image can feed the block after it. Follow the lanes — that's your work, connected.

  1. 1

    Place

    Drop a block onto the canvas and write its prompt.

  2. 2

    Connect

    Mark a block as a source; its output streams down a lane.

  3. 3

    Run

    Hit play — each block pulls from the one upstream.

#1
Launch caption
gpt-4.1

Prompt:

Write a warm, inviting 3-line Instagram caption announcing our new single-origin, Ethiopia Yirgacheffe — bright and floral.

Output:

SNS / InstagramAuto
Spring's first pour is here ☕️ Ethiopia Yirgacheffe — jasmine on the nose, bright citrus on the finish. Come meet your new morning. #AoyamaCoffee #SingleOrigin #Yirgacheffe
#2
Product hero
Flux.2

est. credit: 1,240

PositiveNegativeAspectFile InputSettings
Product photography: a kraft bag of Aoyama single-origin coffee on a sunlit café table, steam rising from a ceramic cup, warm morning light.
1:14:59:1616:9

Output

Product hero — generated image 1Product hero — generated image 2
#3
Story variation
WAN 2.2

est. credit: 2,980

PositiveNegativeAspectFile InputSettings
Recompose the same coffee-bag café scene as a vertical phone story, with extra empty space at the top for a text overlay.
1:14:59:1616:9

Output

Story variation — generated video 1Story variation — generated image 2
Creative engine

Words become images. And video.

Describe it, generate it, keep what works. Every result lands in your assets gallery — a growing body of on-brand work.

Assets
Generated with Flux.2Generated with WAN 2.2Generated with Qwen-ImageGenerated with HiDreamGenerated with VeoGenerated with Stable Diffusion 3.5Generated with Z-ImageGenerated with Qwen-Image

Image + video

Stills and motion from one block.

Block-to-block

Feed an image into the next render.

Private

Output lives in private storage.

Credit-based

Each generation draws from your monthly credits.

Models

New models, no waiting.

The lineup is curated server-side and data-driven — new models appear without a redeploy. So you're always on something current.

Text

gpt-4.1defaultgpt-5.5gpt-5.2gpt-5.1gpt-5gpt-5-minigpt-4o

Creative · ComfyUI Cloud

Flux.2 sampleFlux.2Qwen-Image sampleQwen-ImageZ-Image sampleZ-ImageStable Diffusion 3.5 sampleStable Diffusion 3.5HiDream sampleHiDreamWAN 2.2videoLTX-2.3videoSeedance 2.0videoLumavideoVeovideo

…and a growing catalog. The list is illustrative — what's enabled is managed server-side.

Production-ready

Everything around the blocks.

The quiet machinery that turns a clever toy into a real production tool.

Web search

Ground a text block in current facts with an optional web search.

Knowledge store

Attach your own material — a vector store grounds output in your world.

Output & export formats

Register a format once (blog, script, SNS) for consistent, clean hand-off.

Projects & assistants

Assistants → projects → blocks. Your work stays structured and retrievable.

NodeConnect

Wire one block's output into the next. Text → visual, visual → visual.

MCU credit billing

One credit unit across text & creative — generations draw from your monthly plan's credits.

Private media storage

Generated media lives in private storage, served via short-lived signed URLs.

Membership

Open by invitation.

metagraphy is invitation-only while we grow carefully. Request an invite and we'll be in touch — bringing on new makers every week.

Request an inviteSee how it works

A single monthly membership, credits included — every generation draws from them. One plan, no tiers, no surprises.

Place your first block today.

Request an invite and start building — we onboard new makers every week.

Request an inviteSee how it works