Summary

  • Futurum launched an API and MCP server on September 10, letting organizations retrieve subscribed research from supported AI tools or their own software.
  • One key serves both interfaces, but not every research entitlement. ETR requires a separate subscription, and source retrieval is not the same as an audit of an agent's final answer.

A research subscription is becoming an input to another company's working environment. Futurum's new API and Model Context Protocol server let customers call its market forecasts, surveys and reports from AI assistants instead of first visiting a research portal. The commercial change is where the material gets used, not the disappearance of the subscription behind it.

Futurum announced availability on September 10. It says access is enabled per organization and restricted at the key level to that organization's own data and subscribed practice areas. The same key covers the conventional API and the MCP server. Its reference to more than 10 million data points describes the scale of the research corpus, not an unconditional promise that every key can retrieve all of it.

The launch also distinguishes specific clients. Claude, ChatGPT, Codex, Google Antigravity, GitHub Copilot and Cursor are in the available group, alongside direct use from software and scripts. Gemini Enterprise and Microsoft 365 Copilot are listed as coming soon. A familiar vendor name is not enough to establish that every product in its family is supported.

The material spans several kinds of evidence. Market models provide sizing and growth by dimensions such as product, use case, industry and region. Decision-maker surveys provide buying intentions and vendor preferences. Reports and Futurum Signal add published research, while account intelligence exposes the customer's own coverage and engagement. Those categories can help answer one question without becoming one interchangeable dataset.

The product page makes a further entitlement explicit: ETR enterprise spending intelligence, including Net Score and Pervasion, requires a separate ETR subscription. That matters because the launch's illustrative analyst workflow combines a market model with an ETR spending-intent trend. It is an example of a possible workflow, not evidence that the base research subscription includes every input in that example.

Interpretation has another boundary. In a May explanation of its survey work, ETR describes Net Score as spending momentum within the citing base and Pervasion as presence in the surveyed sector base. These are different measures; neither should silently become a vendor's revenue growth or revenue market share in a generated answer. The methodological explanation is useful here, not the older article's findings as a new September market reading.

Futurum says retrieved research remains traceable to its published source and that it does not train models on customer usage. It also describes a narrower view of the workflow: it sees which research an organization pulled and when, but not the prompt, model or final work product. These are provider statements, not findings from an independent security test.

That last distinction leaves an important job with the customer. A source attached to retrieved material helps establish provenance. It does not, by itself, reveal whether the final answer combined the same survey wave, market definition and unit of measurement. Nothing in the public announcement establishes an immutable version archive or a complete record of the agent's reasoning and output.

Sources