Eunoia is designed as a shared brain for many products, not as a single chat assistant. That has consequences for how we train it, how we update it, and how we expose it to developers.
Instead of tailoring the base model to one persona, we keep it as neutral as possible and rely on higher-level layers and prompts for product-specific behaviour. This reduces conflicts when different products pull the model in different directions.
For developers, this means the public interface will look more like "create a session with clear structure and tools" than "talk to a personality." The goal is predictability, not charm.