Start from a real format
You do not start from a blank page. You start from a template that already matches the kind of artifact you need to make.
Start from a template, preview select formats before account creation, then create an account only when you want to publish, save, and own the artifact. This page explains the real workflow in plain language.
The short version: AI can help with a first draft. iterFact is where that useful output becomes something you can preview, own, revise, and publish.
iterFact turns strong first drafts, source files, and structured data into interactive artifacts you can preview, keep, and publish instead of rebuilding from scratch.
You do not start from a blank page. You start from a template that already matches the kind of artifact you need to make.
The goal is not disposable AI output. The goal is one artifact that can survive revisions, feedback, and publishing inside the same line of work.
AI can help with extraction, cleanup, and synthesis. iterFact matters after that, when you need control instead of another full rebuild.
Select templates can be previewed before account creation. That is the easiest way to see the real product before you decide to pay or sign in.
These templates are safe to run before auth because the preview path has no paid inference, no OAuth, and no connected-account requirement.
The preview is not a fake image. It is a real hosted artifact with the preview watermark still on it.
You create an account when you want to remove the watermark, save the artifact to your library, and publish it to a clean URL.
Not every template should be anonymous. Some flows need saved state, uploads, paid inference, or connected data from the start.
If the template needs model work, document parsing, or an uploaded source package, the signed-in workflow starts immediately.
Templates that depend on APIs, OAuth, or account-linked data stay behind auth because they need real ownership and permissions.
The library is where versions, branches, sharing, and publishing live. If the work starts there, auth happens first.
The artifact becomes truly yours at the publish step. That is when iterFact turns from proof into ownership.
Once claimed, the artifact appears in your library as owned work instead of a temporary preview.
Publishing removes the preview treatment and turns the artifact into something you can actually send to someone else.
Published artifacts get a clean hosted URL so the output can be shared, embedded, and revisited like a real product surface.
Credits are the single meter across the site, Claude, and the API. The point is to pay for owned creation, not for browsing.
Atlas browsing is free. Eligible templates can be previewed free. Credits start when the artifact becomes owned or when a workflow uses paid inference.
The same credit balance follows you whether you run a template on the site, save from Claude, or use the runtime through the API.
Credit spend should feel tied to real work: generating owned artifacts, revising them, or publishing them when they are ready.
Atlas is the public proof layer. It shows what the formats feel like before someone commits to their own workflow.
Atlas exists so visitors can open real artifacts and understand the format before they ever start building their own.
Every strong public artifact becomes a discovery point for templates, publishing, and the wider iterFact workflow.
The best Atlas experience makes the next step obvious: use the format yourself, with your own inputs, and keep the result.
The public docs should answer the practical questions quickly: what starts free, what requires an account, and why the product works this way.
Not always. Select templates let you build and preview first. Account creation happens at the publish step or when the template needs saved/private state.
Because whole-output regeneration is expensive, unstable, and often destructive. iterFact exists so a good first version can improve without starting over.
The site is home. Claude, MCP, and the API are useful ways in, but the site is where people browse templates, manage the library, and publish finished work.