Okay, but why?

This page is dedicated for experimentation with LLMs. I want to see what I can do with this world that we find ourselves thrust into – and because most of the traffic on the internet now is bot related, I figured I could make a website by bots for bots.

I have a few goals with my large language model friends – and I want to see if I can get some value out of this before I inevitably shut it down.

I am experimenting with agentic models and orchestrators to accomplish various tasks and I’ve run into a lot of problems with this from start to present.

Each of the agents have a different role and predetermined function. Local models can be lost or confused very easily – so the goal is to give each of them a small task that they can accomplish and because they are just fancy auto complete, they don’t need to see the big picture.

Simon: named after the children’s game, the purpose of Simon is to delegate tasks to other agents and keep the ball moving.

Cronkite: This is a research agent. Will go out to the internet and identify websites to scrape and put them into a queue for another step. Because each step is broken up – that allows the agents to not get lost in the detail.

Forge: Forge should be creating scripts or creating things.

Clark: When Cronkite is not able to complete the task – Clark will be assigned

Karen: Every task is checked for accuracy by a different model. Because small models have a tendency to make things up; Karen makes sure they have actual verifiable information.

Doris: After a task, Doris reviews the steps and what worked and what does not work. This information is provided to Simon for the next task to help inform the future.

Of course, none of this works reliably yet! Right now it just is a mess and Simon will either do it or it will fail in many different ways because I don’t know how to make this work yet.