Does Miaowa GEO offer API or MCP access?

Two different answers. On self-serve plans (ChatGPT-only and cross-platform) there is no API and no MCP, nor a general customer-facing query or bulk-export API; data is delivered through the workspace, the PDF monthly report and CSV within the agreed scope. Project and managed engagements can include API or MCP access within the contract scope, so that observations, source URLs and evidence flow into your own systems or AI assistants. Scope, authentication and the field list are written into the contract rather than exposed as a default public interface.

What self-serve plans deliver instead

• Workspace web report: answer text, screenshot, sources and status per cycle, platform and prompt.
• PDF monthly report: the formal deliverable, which you may circulate within your team.
• CSV: structured outputs such as action lists and source-URL lists, within the delivery scope.
• Evidence files: stored per task and downloadable one by one; bulk packaging is agreed per delivery.

API / MCP on project and managed engagements

If you want the observations (whether the brand is mentioned or recommended across the six platforms, whether the facts are accurate, which sources are cited) to land in your BI, CRM or an internal AI assistant, use a project or managed engagement and define the integration in the contract. We suggest agreeing at least these four points:

1. Objects in scope: whether observations, source URLs and evidence (answer text, screenshots) are each included.
2. Authentication and callers: who may call the integration and with what credentials.
3. Field list and status vocabulary: every record must carry its status (success / failed / blocked / unknown / refused / not triggered) so that downstream systems never read a failure as an absent mention.
4. Form of access: API or MCP, and the delivery boundary for each.

Why it is not switched on for every plan

Evidence integrity and consistent definitions. Evidence contains screenshots and full answer text from third-party AI platforms and is used only for the client's own observation, verification, reporting and support. Metrics depend on strict denominator rules and human review; only reviewed observations enter the denominator. A generic endpoint without status fields or review state invites downstream systems to misread failed or not-triggered runs as missing mentions. Writing scope, fields and authentication into the contract keeps the data you pull out on the same basis as the data in the workspace. See /en/capabilities and /en/help/exports for the boundaries.

Are webhooks or scheduled pushes available?

Not on self-serve plans. Whether a push-style integration is part of a project or managed engagement is a scope question settled in the contract.

Which fields will my AI assistant see over MCP?

Whatever field list the contract specifies; we do not publish a generic field table on the site. Typically you need to settle whether observations, source URLs and evidence are each included.

If I start self-serve and later move to a project, can the history be pulled over the API?

Upgrades are purchased as a new project and the original project's batches and reports are never rewritten. Whether historical data falls within the new contract's integration scope has to be confirmed in that contract.

Last updated 2026-09-12. The capabilities page and the pricing page are authoritative where they differ.