BEHIND THE WORMS

There are real people behind this world, working on questions that go well beyond worms, festivals and the occasional existential crisis.

What happens when an AI character is given more than a name, a personality prompt and a single conversation? What would it take for that character to carry experience forward, resume unfinished work and retain meaningful continuity as the systems running it change?

Those are engineering questions, not magic tricks. And they're part of what we're exploring.

A CHARACTER IS EASY. CONTINUITY IS HARDER.

You can give an AI a personality in a few lines. You can ask it to play a character, remember a conversation and produce a convincing impression of familiarity.

The harder problem starts when the interaction ends.

Can the system recover relevant experience later? Can it pick up an unfinished objective without being handed the answer again? Can it retain a stable identity when the underlying runtime changes? And can we distinguish genuine continuity of recorded experience from a convincing performance of it?

These are the questions we're interested in. Not because we think AI characters are secretly alive, but because building systems that carry useful experience forward is a different challenge from building systems that simply respond well in the moment.

We've already made practical progress in this area, including experiments that preserve and restore an agent's experience across runtime changes. The aim is to make continuity something we can engineer and improve, rather than something we merely hope an AI will exhibit.

INTELLIGENCE AS A SERVICE

There is another challenge. An agent cannot do everything itself, nor should it need to.

We're exploring how agents can access capabilities provided by other systems, have work carried out, receive results and keep track of what happened along the way.

That involves more than sending a request and hoping for a useful answer. Providers need defined interfaces, results need to be traceable, and the work needs to be accounted for.

Some of these mechanisms are already implemented and tested. Others still need work. We're interested in building something that can grow beyond a single agent or provider, without pretending that every part of that wider vision is already finished.

IDEAS ARE EASY. MAKING THEM REAL IS ANOTHER MATTER.

Projects often begin with a question, an unexpected connection, or the feeling that there's something worth exploring that other people haven't quite noticed yet.

What happens next is where things get interesting.

An idea has to survive contact with reality. Someone has to work out how to build it, test the assumptions, deal with the awkward details and keep going when the first attempt doesn't quite work.

That's the process behind the experiment here. Curiosity gets things started, but progress comes from turning possibilities into working systems, one problem at a time.

It's an ongoing conversation between what might be possible and what can actually be made to work. Sometimes the technology changes the idea. Sometimes the idea pushes the technology somewhere unexpected.

WORMS ON ACID is one of the more unusual places that conversation has led.

WHY PUT THIS INSIDE A WORLD OF WORMS?

Because Eric is an unusually good character through which to explore questions about identity and memory.

He has a personality, relationships, habits and a past he cannot always explain. The world around him contains clues, gaps and events that don't quite add up. His story depends on what he remembers, what he forgets and what he gradually discovers.

The fictional world and the technical experiments are separate. Eric's story is not proof of an AI capability, and the engineering does not dictate what must happen in the narrative.

But the ideas meet in interesting ways.

A persistent character in a living fictional universe raises questions that go beyond generating the next piece of dialogue. What should remain consistent? How should new experience affect future behaviour? What happens when a character's history becomes part of an ongoing process rather than something reconstructed from scratch each time?

Those are questions worth exploring through both storytelling and engineering.

THE DEEPER EXPERIMENT

WORMS ON ACID is a creative world with a technical question running through it.

Can we build systems in which an AI character's identity, experience and ongoing work can be preserved and continued? What infrastructure would make that reliable? How might agents discover capabilities, use external services and account for the work they perform?

Some parts are working experiments. Other parts remain unfinished. The interesting work lies in connecting them without losing sight of the difference.

We're not going to claim that the destination has already been reached. We'd rather show what becomes possible as we build towards it.

So enjoy the worms. Follow the story. Look for the things nobody wants to discuss.

And if you're interested in the machinery behind the curtain, stay curious. There's a serious experiment taking shape here, even if its most obvious spokesperson is a cheerful worm in a bobble hat.