AI Usage
With AI Usage you see in one place everything your assistants, flows, robots, Designer tests and IDP projects ask of a language model, of audio transcription and of scanned-document reading. You know how much is used, who uses it and how much of your plan’s included usage is left, without adding up bills from several providers.
Two ways to use AI, one panel
Section titled “Two ways to use AI, one panel”| Included in the plan | Your own account | |
|---|---|---|
| What you set up | Nothing: leave the credential empty | Your credential (provider key or Azure resource) |
| How it travels | Through Zoan Cloud, with Zoan’s accounts | Straight from your robot to your provider |
| What is recorded | The usage: tokens, audio seconds, pages and their estimated cost | Only the usage, so you see it alongside the rest |
| Counts toward what your plan includes | Yes | No |
You can mix both: one activity with your credential and another without, or your own model with the included document reading.
With the included AI, calls from robots and Designer tests go through Zoan Cloud to the provider (Anthropic, OpenAI or Microsoft Azure), and Zoan Cloud does not store the content of those calls —not the prompt, the answer, the audio or the document—: only the usage. What you store on purpose does stay on the platform, with its retention: IDP documents (the original, its pages and its extraction) and assistant conversations.
Personal-data masking is on by default in the AI and IDP activities (maskPii), and your organization can make it mandatory: text goes to the model with placeholders instead of ID numbers, emails or phones, and the real values come back in the result. Images, scans and audio cannot be masked: a scanned document goes in full to Azure Document Intelligence to be read, and an audio file goes in full to the transcription service.
What you see in the panel
Section titled “What you see in the panel”- Included in the plan and your own account, separately.
- By source: assistants, flows, robots, Designer tests, IDP and knowledge bases.
- By process: which robots use the most, with their Designer tests apart.
- By model: input and output tokens, transcription minutes and pages read.
- What is left this month: the included AI usage and the document page quota.
What counts
Section titled “What counts”| Usage | How it is measured |
|---|---|
| Generating, extracting and classifying with a model | Input and output tokens of each call |
| Transcribing audio | The duration the transcription service reports; if it does not report it, it is measured from the file (exact for an uncompressed WAV) or estimated from its size |
| Reading scanned documents | Pages read by Azure Document Intelligence (with your own resource they are measured but don’t count toward the plan’s quota) |
| Pages stored in Zoan Cloud | Every page of a document stored for the validation station, digital or scanned. IDP Extract stores the document when it extracts it: a digital PDF pays no reading, but its pages do count toward the quota |
For transcription, Whisper models report the duration; with gpt-4o-*-transcribe it is measured from a WAV or estimated from the size. Designer tests count too: they are real calls to the provider.
Reaching the monthly limit
Section titled “Reaching the monthly limit”What your plan includes has a monthly AI usage limit and a document page limit. When you reach it, calls with Zoan’s account stop with a clear message and resume next month or when you upgrade the plan. AI and reading with your own credentials keep going. Pages you store for the validation station do count toward the plan’s page quota.
With documents, Zoan Cloud gets ahead of it: if a document does not fit in what is left of the quota, it is rejected before it is read and you get the numbers, instead of paying for a reading you cannot use. And you never get a partial reading presented as complete: if the number of pages cannot be known in advance and the reading fills the quota, the result is not delivered and the activity fails with a clear error.
| Message | What to do |
|---|---|
| The AI usage included in your plan for this month has been reached | Use your own credential in the activity or upgrade the plan. |
| The document pages included in your plan for this month have been reached | Use your own Azure Document Intelligence resource or upgrade the plan. |
| The document has N pages and M are left in your plan’s quota this month | Read fewer pages (pages parameter), use your Azure resource or upgrade the plan. |
| The reading reached N pages… and the document may have more | The reading filled the page quota or the included AI usage and was not delivered so you do not get an incomplete result. Split the document with pages or use your Azure resource. |
| The document has N pages and one reading takes up to 2000 | Read it in parts with pages. |
Robots show these messages in their language (English or Spanish).
Next steps
Section titled “Next steps”- AI Complete, AI Extract and AI Transcribe — AI in robots.
- IDP Load Document — reading documents with the included reading.
- Document processing (IDP) — extraction projects without credentials.