Platform / Adaptive agents ISABELLA AI
A local-first research and engineering platform connecting bounded perception, agent-scoped memory, selective reasoning, and structured action.
Explore the platform →Portfolio / Research in context
We give intelligent systems somewhere real to prove themselves.
Our portfolio is built around internal proving grounds: environments where memory, context, structured action, and integration have to work together. They expose the difficult engineering questions that a presentation-only demo can hide.
Selected internal work
Platforms designed to reveal the hard parts.
These projects demonstrate WPS research direction and integration practice. They are not presented as customer deployments, third-party validation, or guaranteed performance.
Platform / Adaptive agents 
Persistent world / Internal proving ground Medieval Life: Chronicles of Annora
A demanding world in development where characters, relationships, resources, events, and simulation state create consequences over time.
Enter the proving ground →What the work tests
Architecture has to survive contact with a product.
Internal projects let us examine the seams between models and software: how context enters, how memory is owned, how decisions become valid actions, and how behavior remains observable.
- Agent-specific knowledge and experience instead of universal context
- World and application events routed to the participants that need them
- Actions constrained by the host product's state, permissions, and rules
- Workload and model choices evaluated inside representative environments
- Human operators and domain experts kept inside the validation loop
Your project should have its own proving ground.
We can define the smallest representative environment that will expose the technical risk and give your team evidence for the next decision.