Utsav A.

Utsav A.

Senior SWE & AI Agent Researcher

Forums

What's really pulling back the progress in agentic AI ?

Now, I know the progress in the field of agentic ai has just really taken a pace. With the introduction of mcp, a2a protocol & anp protocols, it certainly made us ask the right question: What really is required for taking a giant leap in the progress of ai agents? Obviously, the answer lies in the question: "quick progress"! Sounds generic right? but think for a second. Nobody can give objective answer to that question, but every answer is what the progress follows.

So to pace up this progress as fast as possible, human intervention needs to be at the absolute minimum. Because let's be honest, humans aren't the most efficient being in this universe, we created machines to cope up against our weakness. So no bad feelings there :)... Anyways, for agents to progress quickly, they need very smooth communications between each other. The better the articulation in their communication, the faster the progress iteration. Now, i'm not claiming to have found the breakthrough to this problem. But i have slightly started answering to the question.

Utsav A.

4d ago

Sockridge - Agent Discovery Infrastructure

What’s different here is the focus on agent-to-agent discoverability as a first-class problem, rather than an extension of existing service discovery or API management tools. Most current solutions assume stable endpoints and human-managed integrations. Sockridge assumes dynamic, autonomous agents that need to find each other based on capability, not location. The combination of semantic discovery, mutual access agreements, and a registry that stays out of the data path is a notable shift.