
Graeme Scott sits down with David Mooter, analyst at Forrester covering integration and APIs, and more recently quantum computing. Two subjects that sound unrelated and turn out to be the two things most likely to determine whether anything you build actually works.
His opening argument is blunt. If your agent cannot reach your data and, more importantly, your business capabilities, the things that actually take action like approving a claim or sending a document, then it is nothing more than ChatGPT version one. Agents run on integration. Without it you have a glorified chatbot with a budget attached.
The uncomfortable part is what that means for API estates that were never built properly. The accepted best practice has always been to design APIs around business capabilities and value streams rather than database operations and low level point to point connections. Plenty of organisations cut that corner, and for years it did not matter much, because there was a workaround. You walked over to the API owner's desk and worked it out between you.
That workaround has just disappeared. An agent will not walk over to your desk. Turn a well designed capability into an MCP tool and it works. Turn a pile of low level operations into MCP tools and the agent has to assemble them itself, which means more tools, more context window consumed, more money spent and worse answers. The same applies to documentation nobody wrote, because there is no longer anyone to ask.
He is equally direct about governance, and about what changed this year. Last year the instruction was to do AI. This year organisations noticed the bill climbing with no way to attribute it to anyone. He describes a company that budgeted one percent of revenue for AI spend, watched it track towards three percent, and had no visibility into who was spending what. Governance is now the single most common question he gets from clients, which he points out was not true at all eighteen months ago.
Then the conversation turns to quantum, which he picked up from Brian Hopkins and has clearly loved. He gives the best explanation of superposition anyone has managed on this show, using a plucked guitar string and its overtones, and explains entanglement with two billiard balls. He is also careful about the hype. These are specialised machines, not better computers. If a classical computer already solves your problem well, quantum will not solve it better.
More episodes from "GAEA Talks"



Don't miss an episode of “GAEA Talks” and subscribe to it in the GetPodcast app.








