We would like to look at agentic systems in this year's People Tech Maritime Athens.

​

Rather than discuss what large language models can do, we should discuss specific tasks and what specific kind of language model is best suited for them.

​

Large language models = expensive, better for complex tasks, you have no control

Small language models = you can run them yourself, you have complete control, they can be provided as part of a software package, but may ultimately find them limiting

​

An agentic network may include multiple small language models co-ordinated by a large language model. 

​

Many people in enterprise technology believe this is where things are going. Is this the pathway for maritime? 

​​

​