
Sounds cool, doesn’t it? It is cool. But it’s also a warning sign.
That’s because Agents are the “next big thing” in AI, after Large Language Models, which have failed to live up to their billing. Making LLMs bigger doesn’t make them smarter.
Agents, which are software brokers, are supposed to make up for this by using a variety of resources to accomplish their tasks. Techies call this orchestration. Agents are supposed to manage your life the way a personal secretary might or manage corporate resources the way a junior executive would.
In practice, they can help both these people, once people are trained to use them, and this could mean some underlings go away. But you’re just upgrading the assistant’s skills, which technology has been doing since Visicalc. That’s good, but it’s not the revolution AI boosters promised. It’s certainly not superintelligence. It’s just software.
What’s So Bad?

The white heat of AI born of ChatGPT was based on the idea that proprietary advances would provide proprietary advantages. The greed exhibited by LLM producers in the wake of the November 30, 2022 announcement was breathtaking. It set off an explosion of investment that continues to this day.
But now all these folks are coming together, admitting that the assumptions behind their proprietary push were false. It’s also important to realize that the AgentGateway will be real open source, not the fake kind pushed by Meta’s LLaMa, which apparently fooled no one.
If the proprietary assumptions behind LLMs were wrong, however, what else was wrong? The industry has come to realize that a lot is wrong. They’re coming together under the Linux umbrella the way golfers flee a thunderstorm for the clubhouse. But this may not be a thunderstorm. It may be a hurricane, with tornadoes, the kind that can rip the roof off that comfy clubhouse and expose everyone.
One More Thing…

AI doesn’t have to fall to cause a panic. Current valuations are so high that it must only slow down to send investors rushing for the exits.
It’s slowing down.







