You know the routine: The IT industry, driven as much by vendor ambition as by necessity, develops many competing standards to solve a simple problem. Today’s culprit: agent-to-agent communication in AI.
The recent rise of so-called “standards” for how intelligent agents should communicate echoes past issues with service-oriented architecture, web services, and various messaging middleware conflicts. The key difference is that now, this confusion could prevent one of the most promising areas in enterprise technology—agentic AI—from ever providing real value.
Let’s set the scene. Intelligent agents, whether they are specialized large language models (LLMs), service-brokering bots, Internet of Things digital twins, or workflow managers, need to communicate efficiently, securely, and transparently. This is a typical interoperability issue. A well-established industry could, in theory, create a straightforward, practical protocol and move forward. Instead, we see a flood of emerging standards from too many “expert” voices with an underlying agenda, each accompanied by a white paper, a community call, a sponsored conference, and, of course, an ecosystem. This is the core problem.