Built, or being built.

Dream interpretation has always lived in two unsatisfying places: pop psychology that flattens it into clichés, and clinical analysis that's expensive and slow. Modern AI gets close to a third option — but consumer AI dream apps either reduce everything to a single LLM prompt, or treat user dreams as training data.

ANIMUS is the version we wanted to exist: a multi-agent AI system trained on serious interpretive frameworks, on infrastructure where the user owns their dream record by default.

A single LLM doesn't interpret dreams well. The interpretive process is intrinsically multi-perspectival — symbolic, emotional, narrative, biographical. ANIMUS uses multi-agent orchestration: each agent specializes in one interpretive lens, agents exchange context, and a coordinator agent synthesizes the final reading.

Privacy was a hard requirement, not a feature. Dream content is among the most personal data a user can produce. ANIMUS runs on a decentralized storage layer so dream records and interpretations live under the user's keys, not in our database. The blockchain part isn't ornamental — it's the only architecture that delivered the privacy guarantees we wanted.