Post

I’m dipping my toes into this water, but with respect to agent orchestration frameworks there are aspects of process automation - for example a planner may decide which agents to call - and some implementations lean towards actor models - agent has inbox, knows about other agents.

I would be cautious around a “just microservices” model, as a microservice could be use Agentic AI, “under the hood”, in the same way it could own a workflow

But, it’s piqued my interest because it’s certainly entering “my wheelhouse” of services, reactive, events, actor models etc. So, I think as a “shorthand”, yes, all the same questions, tools etc.

Agents seem to imply some use of AI vs hardcoded behaviour?

That's it... You don't "program them", you explain to them in English what they are, where they read and write the information... But they do the same thing as hardcoded dockers

Well “program” an agent in the sense that you wire up inputs to some business logic that spits an output. The core diff is that said business logic is implemented via calling an AI model vs a coded procedure (e.g. actual code) Both are the same in the sense that they turn inputs into outputs ->