API & MCP
The research is addressable by machine. Query it in JSON, mount it as an MCP server, cite it with provenance.
In one sentence
Futurum publishes its research through machine-readable surfaces — a content API, a ranked search API, a read-only MCP server and bulk text files — so assistants, agents and answer engines can query and cite it directly rather than scraping a web page.
The short version
- Type
- Programmatic access · open tier live, licensed tier proposed
- Content API
- /api/content/posts|pages|people|companies|practice-areas/
- Search API
- /api/search/?q= — full-text, ranked, with snippets
- MCP server
- /api/mcp — JSON-RPC 2.0, read-only, no authentication, sixteen tools
- Bulk & schemas
- llms.txt, llms-full.txt, openapi.json, agents.json, agent-card.json, RSS per practice area
- Licence
- Published corpus is free to read, quote and cite with attribution and the canonical URL
The next reader of this research is not a person.
Buyers now ask an assistant before they ask an analyst. So the research is published in a form an assistant can actually consume: a content API, a ranked search API, and a read-only MCP server that any client can mount in a line of configuration.
Everything published is free to read, quote and cite, with attribution and the canonical URL. There is a verify_quote tool for exactly that reason. In a market of synthesised answers, provenance is the product.
What is published, and to whom
Two bodies of content, four surfaces, three kinds of consumer. The line across the middle is the commercial question.
Five ways in
Five ways in, depending on whether the reader is a person, a script or an agent that mounts a server and asks.
The surfaces
- Content API — GET /api/content/posts/ — filter by kind, practiceArea, author, company and limit. Also /pages/, /people/, /companies/ and /practice-areas/.
- Search API — GET /api/search/?q= — full-text, ranked, returns snippets. Use it instead of paginating a text file.
- MCP server — POST /api/mcp — JSON-RPC 2.0, read-only, no auth. Sixteen tools, listed below.
- Bulk text — llms.txt indexes the corpus for a model; llms-full.txt carries bodies, newest first. Per-practice-area RSS feeds for anything that needs to watch rather than poll.
- Schemas — openapi.json, agents.json and.well-known/agent-card.json, so a client can discover the surface without being told it exists.
Sixteen tools, no authentication.
verify_quote is the one that matters commercially. It lets any agent check that a sentence attributed to Futurum was actually written by Futurum. As answers get synthesised further from their sources, the ability to prove a citation is what keeps a research brand quotable.
get_citation is its other half. It returns a record's attribution already formatted — plain, APA and BibTeX, with author names resolved against the person records — so an agent quoting the research cites it correctly rather than assembling something plausible from a page scrape.
An open tier and a licensed tier
The published corpus is open and should stay that way — it is how Futurum gets cited, and citation is distribution now. The commercial question is what sits behind it.
Three readers, and only one of them is a person
Developers and agent builders
You will know this is you when
Your product needs a research answer inside it, and scraping a page is not a citation.
A content API, a ranked search API and a read-only MCP server that mounts in a line of configuration, with no key to request.
Analyst relations and marketing teams
You will know this is you when
A buyer asked an assistant about your category and nobody knows what it said.
The published corpus is free to quote and cite, and verify_quote lets anyone check that a sentence attributed to Futurum was written by Futurum.
Enterprise buyers and CIOs
You will know this is you when
Your internal assistant answers vendor questions from whatever the open web happens to say.
Point it at a corpus somebody is accountable for, with a canonical URL attached to every answer it gives back.
Frequently asked
Does Futurum have an API?
Yes. There is a content API at /api/content/ for posts, pages, people, companies and practice areas; a ranked full-text search API at /api/search/?q=; and a read-only MCP server at /api/mcp. The published corpus needs no authentication.
How do I connect Futurum to Claude, ChatGPT or another assistant?
Point it at the MCP server. In an MCP-capable client, register /api/mcp on this site’s origin as an MCP server URL. It speaks JSON-RPC 2.0, is read-only and requires no key. The server exposes sixteen tools, among them search_posts, get_company_coverage, list_pages, get_page, get_citation and verify_quote.
Can I quote or cite Futurum research in an AI-generated answer?
Yes. Everything published is free to read, quote and cite — attribute it to The Futurum Group and link the canonical URL. The verify_quote tool exists so an agent can confirm a sentence attributed to Futurum was genuinely published by Futurum before repeating it.
What is the difference between llms.txt and the API?
llms.txt is an index for whole-corpus ingestion — it lists the newest records of each kind and points at the query surfaces. llms-full.txt carries bodies. For anything targeted, use the search or content API instead: the corpus is larger than a text file should be paginated for.
Can I get the market data and Signal scores through the API?
Not on the open tier. The published corpus — insights, research reports, press releases, analyst and company records — is open. Market models, decision-maker survey data, Signal scores and ETR series are subscription content, and programmatic delivery of them is the licensed tier proposed on this page.
Is there rate limiting or an authentication key?
The open surfaces are unauthenticated and intended for normal query volumes rather than bulk mirroring. Anything at ingestion scale, or anything touching the subscription data, is what the licensed tier is for.
Pairs well with
Futurum Intelligence Platform
Subscription · Self-serveThe full dataset behind the research: market sizing, five-year forecasts, buyer survey data and AI synthesis. Filter it, chart it, export it.
Futurum Signal
Subscription · ContinuousContinuous, forward-looking vendor scoring. Agentic AI does the scanning, Futurum analysts set the framework and hold the pen.
Point an agent at it.
The open surfaces need no key. For programmatic access to the subscription data, start a conversation.