Fascinating to see how fast LLMs keep following what looks like an evolutionary trajectory.
Since 2022 we've moved from raw cognitive muscle to something structured, collaborative, tooled, and now surfacing into interfaces. For months now, several of our digital teams have been enthusiastically exploring AG-UI, which brings a native front-layer of interactivity to our agentic applications. With the newly proposed "MCP Apps" extension, the creators of the MCP protocol are clearly acknowledging this emerging need in these systems. For sure the coming months will be fascinating as the ecosystem of interoperability protocols continues to take shape.
But it also leads me to a broader—and more darwinian—reflection.
In A Brief History of Intelligence, Max Bennett highlights the remarkable parallels between the major evolutionary leaps of biological organisms—steering, reinforcing, simulating, mentalizing, speech—and the progressive phases of information-processing systems today. Looking back at the post-ChatGPT era, the first months really felt like a Cambrian explosion of models. But what I find striking is how darwinian the trajectory has been: the early large models, prone to hallucinations and black-box reasoning, quickly required the externalization of structure—chain-of-thought, RAG grounding, mixture-of-experts architectures—functioning almost like an exoskeleton of guardrails, just as arthropods first solved structural weakness through external shells.
Then, as models left the water and entered real production environments, the evolutionary pressure shifted: tools emerged—MCPs, orchestrators, agentic patterns—exactly as organisms developed manipulable structures enabling new behaviors.
And now, with standards like AG-UI, MCP-UI and that MCP Apps extension of the protocol, these systems are surfacing into interfaces: developing the skin and sensory layer through which cognition finally meets interaction.
In the end, it only reinforces how our digital systems are no longer truly designed—they are grown, shaped by pressures, constraints, and adaptations we discover along the way.
