For creators, agencies & studios
Save every workflow. Deliver work clients can verify.
Connect Claude or ComfyUI once and Archibal records every generation: prompts, models, seeds, and references. Your work becomes a searchable library where nothing is lost - and every delivery ships with a signed provenance record your clients (and their platforms) can verify.
The Full Workflow, at a Glance
From the first capture to a delivery your client can check - five steps, mostly automatic.
Set Up a Project
A workspace per job. Create the project, note where the work will run, and every render you make lands here with its context attached.
Provenance tool
Sign any standalone asset with Archibal's C2PA identity and mint a public verification link - no project needed. Also available via the v1 API and ComfyUI node.
How your record earns trust
Every claim carries its evidence grade - attested, integration-reported, studio-logged, or user-declared. See how verification works.
Read the docs
The Creator guide covers the full lifecycle: project setup, model and prompt declarations, and delivery.
From the blog
Frequently asked questions
Do I need to change my production workflow?
No. Connect ComfyUI or Claude once, or upload finished assets; Archibal records the models, prompts and workflows behind the work and signs the delivery. Your tools stay your tools.
What does my client actually receive?
A signed AI provenance record with a public verification page and C2PA-signed manifests for every asset - documenting exactly which models, prompts, and reference material made the work.
Does an Archibal record satisfy EU AI Act Article 50 for me?
Article 50(2) machine-readable marking binds AI model providers, and Article 50(4) audience disclosure binds deployers - not the agency directly. What the record gives you is the cryptographic provenance documentation (the technique Recital 133 names) that brands and platforms increasingly demand from every agency.
What about the models' own terms - can I use the output commercially?
Every model route in your project carries its usage terms on the record: whether the provider allows commercial use, who owns the output, whether your prompts are used for training - both for the channel you used (fal, Higgsfield, Comfy Cloud) and for the model's own provider. Sourced from the providers' documents, shown plainly, and frozen into the delivery.
Commissioning AI work instead of producing it? See how Archibal works for brands