with Ed Anuff and Anant Jhingran
MCP-ization and GraphQL: Enabling AI-Driven Open Data Access
July 31st, 2025 | 58:48 Runtime
Episode Guest
Roy Derks
Principal Product Manager, IBM
Episode Transcript
TIMESTAMPS
[01:18] – MCP’s growing centrality in all agentic work.
[04:58] – Ed on blending human strategy discussions with LLM insights.
[17:55] – Roy explains interoperability issues across agent frameworks.
[30:13] – Roy explains GraphQL’s ongoing relevance for agents.
[52:26] – Ed highlights why GraphQL is a powerful fit for agent architectures.
KEY QUOTES
"But as you and I have talked about many times at the moment, all paths seem to lead through MCP to any interesting thing you're doing." —Anant Jhingran
"I think ultimately there's a synthesis of that and I really do think that a lot of this is about building the user experiences for actually not having to do those in two tracks." —Ed Anuff
"And then this makes it kind of hard to bring your code from one agent framework to the other. So this led to issues. And then it also makes it hard for vendors or for people building dev tools to create tools that were easy to use and reuse." —Roy Derks
"GraphQL, is very powerful technology within the API space. That in seeing Roy's demo and so on, in fact, where I would actually think that we would even be able to go deeper is a technology that is uniquely suited for going and dealing with the types of problems which are entity relationships and really traversing and finding the right piece of information that are going to be directly related to the types of, you know, agentic tasks that we're giving that we're requesting the agent to do." —Ed Anuff
"So if developers can use this, then agents can use this as well. Because most modern LLMs, they have been trained on code bases which include GraphQL, so they're able to parse, natural language to GraphQL and the other way around." —Roy Derks
[01:18] – MCP’s growing centrality in all agentic work.
[04:58] – Ed on blending human strategy discussions with LLM insights.
[17:55] – Roy explains interoperability issues across agent frameworks.
[30:13] – Roy explains GraphQL’s ongoing relevance for agents.
[52:26] – Ed highlights why GraphQL is a powerful fit for agent architectures.
KEY QUOTES
"But as you and I have talked about many times at the moment, all paths seem to lead through MCP to any interesting thing you're doing." —Anant Jhingran
"I think ultimately there's a synthesis of that and I really do think that a lot of this is about building the user experiences for actually not having to do those in two tracks." —Ed Anuff
"And then this makes it kind of hard to bring your code from one agent framework to the other. So this led to issues. And then it also makes it hard for vendors or for people building dev tools to create tools that were easy to use and reuse." —Roy Derks
"GraphQL, is very powerful technology within the API space. That in seeing Roy's demo and so on, in fact, where I would actually think that we would even be able to go deeper is a technology that is uniquely suited for going and dealing with the types of problems which are entity relationships and really traversing and finding the right piece of information that are going to be directly related to the types of, you know, agentic tasks that we're giving that we're requesting the agent to do." —Ed Anuff
"So if developers can use this, then agents can use this as well. Because most modern LLMs, they have been trained on code bases which include GraphQL, so they're able to parse, natural language to GraphQL and the other way around." —Roy Derks