SPONSORED: Patricia León Elías, senior product marketing manager at Reuters News Agency, explains its latest foray into AI tech
For decades, getting trusted news content into a workflow meant an API integration: development, technical resource, and a system built to be read by other systems. That was fine when the only thing on the other end of the pipe was software written by engineers.
That’s no longer the only case. AI agents are now part of the workflow too, sitting inside newsroom tools, research platforms and production pipelines, expected to pull in accurate, licensed content the same way a producer would search an archive. Reuters started 2026 by setting out to close that gap.
The problem with reworking old systems
Most of the industry’s response to agentic AI has been to rework access onto what already existed, wrap an old API in a chat interface, or point a model at a search index and hope the licensing holds up. It works, up to a point. But it treats AI access as an afterthought, and in a newsroom or production environment, this workaround isn’t good enough when the content needs to be both fast and right.
Reuters had heard this challenge repeatedly from customers, including at Reuters Future of News events globally. So, we took a different starting position: build for AI-native access from the ground up, rather than retrofit it.
What it actually changes
In practice, MCP lets a customer’s AI platform query Reuters text, photos and video directly, in natural language, instead of writing code against a traditional API. Setup takes days rather than months. Licensing and entitlement controls apply automatically, so an AI system only ever surfaces content the customer is already subscribed to.
Reuters tested this with newsrooms before it ever launched. In a five-customer trial earlier this year, we had trialists that integrated Reuters content into its newsroom research tool in about a day. Another ran text, image and video retrieval in a non-English language as part of its own AI workflow. A third focused on image search and media discovery. The common thread: fast integration, and confidence that what came back was Reuters-verified.
Alphonse Hardel, head of Reuters News Agency, explained: “As organizations increasingly rely on AI to process information, demand is shifting toward AI-native ways for our customers to access Reuters content. With the Reuters MCP server, we’re at the forefront of making our trusted content accessible inside their AI platforms.”
Why this matters for broadcasters
None of this changes how Reuters journalism gets reported, verified or edited. What it changes is how quickly a broadcaster’s own technical and editorial teams can put content to work inside whatever AI tools they’re already building or buying.
As newsrooms and production facilities move more of their workflow into AI-native tools, research, verification, rights-cleared content discovery, agentic publishing, the content layer underneath needs to keep up. MCP is Reuters’ answer to that shift, and it’s only the first step: support for more content types and deeper agentic capability is already on the roadmap.
See it live at IBC
The best way to understand this is to see it working. At IBC2026 in Amsterdam (11–14 September), Reuters will be running live demonstrations of the MCP server on the stand (Hall 5), showing exactly how a natural-language query pulls verified Reuters content straight into an AI pipeline. No integration time required.
Come and see it for yourself and talk to our product and technology team about what it could mean for your newsroom.
