One subscription. The AI layer only if you want it.
The platform is a flat fee and runs on your own model provider keys. If you would rather not hold keys at all, switch on the UEMA AI layer — billed separately, metered per token, invoiced monthly, and off again whenever you like.
Optional. Run the judgment steps on UEMA's own inference instead of your provider keys — metered per token and invoiced separately, monthly.
Platform
$1,200
per workspace / month
The full accelerator, running on your own model provider keys. One subscription, unlimited pages, unlimited migrations.
- A rendered-HTML crawl that works against any CMS, plus native connectors where they help
- The full exemplar loop: region detection, clustering, block mapping and importer authoring
- Deterministic bulk migration over unlimited pages, at zero inference cost
- Both AEM targets from one block definition — Edge Delivery and the Universal Editor
- Block synthesis: block.js, block.css, UE models, authoring tables and component definitions
- Eight blocking verification gates, and a review queue that records every decision
- Real emission, a real diff against your branch, and a real pull request
- Accounts, tenants and RBAC, with tenant-scoped projects
- Thirteen AI providers behind one setting, plus any OpenAI-compatible endpoint — including your own
- Per-token usage and cost tracking, whichever keys you are using
- Unlimited projects, unlimited seats, unlimited migrations
Two lines on the invoice, at most
1 · Platform
Predictable and unmetered. It does not move with page count, project count, seat count or how much you migrate in a given month — the bulk pipeline is deterministic, so there is nothing there to meter.
Includes bring-your-own-key across thirteen AI providers. Your provider bills you directly for the exemplar passes, and UEMA tracks that spend per token too.
2 · UEMA AI — optional
Switched on from the panel above, and it appears as its own line. Every token in and out is counted against a cap you set, shown live in the product, and invoiced in arrears at the end of the month.
Switch it off and the line disappears from the next invoice. Your platform subscription is unaffected, and the product keeps working on your own keys.
The Claude Agent
A standalone migration agent, licensed separately from the platform. It runs on your own machine, in your own Claude session, against your own repository.
What the licence covers
- Unlimited migrations and unlimited pages for the term of the licence
- Both Adobe targets — Edge Delivery Services and AEM Cloud Service
- Runs inside Claude, so no separate AI account and no second usage bill
- Reads any CMS from the rendered page, and directly where the CMS allows it
- Components, content, images, navigation and site configuration, in one project
- Every page checked against the original, with a report a content owner can read
- A demo against a few pages of your own site before you commit
- Onboarding for your team, and a named contact while the migration is running
Priced per engagement
The sites it is bought for range from a few hundred pages to a few hundred thousand, so a list price would be wrong for almost everyone who read it. Tell us roughly what you have and you will get a real number.
Contact salesWhat it doesYou see the tokens before you see the invoice
Usage is broken out by role and by model — labelling, mapping, importer authoring — with input and output counted separately, because they are priced separately.
- A monthly spend cap you set, enforced before a request leaves rather than reconciled after
- The same view whether you are on your keys or ours — the tracking is not a billing feature
- No charge at all for the bulk pass, because it makes no model call
Usage & billing
UEMA AI · metered per token · billing period 1–30 September
$21.23 this month
Spend cap $250.00
Enforced before a request leaves — a run that would exceed the cap is refused, not invoiced.
Invoiced monthly in arrears, and only while the AI layer is switched on. Bulk migration appears nowhere here — it makes no model call.
What else you pay for, and who to
Your own model provider
On the judgment steps only — one to three exemplar pages per pattern, never one call per page. You hold the keys and the provider bills you directly.
Whatever you run it on
UEMA is self-hosted on infrastructure you control. A single small instance and a managed Postgres is enough to start.
Switching on the UEMA AI layer removes the first of these. The second stays either way — UEMA runs on infrastructure you control.
The ones people actually ask
Anything not covered here — procurement, security review, a longer term — is a conversation rather than a page.
Talk to usBecause the model does not see most of your pages. It migrates one to three exemplars interactively, a human approves the result, and the agent then writes a deterministic TypeScript importer module with fixture tests. That module runs over the rest — in parallel, resumable, at zero model cost. A 100,000-page site costs roughly what a 100-page site costs in inference.
See it against your own site first
A working session on one of your real templates tells you more than any pricing page. That is usually where this starts.