An interior-state architecture for the products you build.

Coherent inner life is not a longer context window.

A robot, a companion device, a wellness platform, a conversational product, each of these increasingly needs a coherent inner life. Today, builders reach for the closest available substitute: prompt scaffolding, short-term memory buffers, sentiment classifiers stacked together. The result is plausible turn-to-turn behaviour with structural inconsistencies at scale. The fiftieth conversation is not informed by the forty-nine that came before it.

The missing layer is not a model. It is a theory. Freud for Machines is that theory, returned to Freud directly, recovered from a secondary literature that had mystified it, and rendered as a computational ontology. Phi-Yo is the first product that resolution made possible.

The product, and the foundation underneath.

Phi-Yo is a computational psychoanalytic wellbeing companion. It runs as a live consumer product at phi-yo.app and as an integration-ready backend for third-party platforms. Phi-Yo v2 has been in production since 26 April 2026, with a native mobile release (iOS and Android) following in May 2026.

Freud for Machines is the formal ontology beneath the product: concepts, premises, propositions, and transpositions that constrain every interpretation Phi-Yo produces. The framework is what makes outputs structurally consistent across modes, across users, and across time. It is not separately licensable; it is available exclusively through Phi-Yo integration.

Each mode addresses a distinct moment in the user's day.

Each mode is exposed as an integration endpoint, and each draws from a single shared profile rather than building its own.

The continuity layer no host could practically build alone.

A single per-user profile underwrites every mode. Four components, all longitudinal:

Context (life situation, current concerns), affective history (recurring emotional patterns and their evolution), personal symbol map (a user-specific map of recurring images with their frequency, contexts, and evolving meaning), and a monthly pattern report synthesizing all four modes.

This is the integration's most valuable asset. Any host platform that integrates Phi-Yo inherits a continuity layer that cannot be replicated through longer context windows or ad-hoc memory buffers. Phi-Yo also generates natively in nine languages (English, French, German, Spanish, Japanese, Russian, Arabic, Chinese, Turkish), with automatic detection.

A small public surface, by design.

Partners receive four endpoints, a memory layer, and the framework that makes both consistent. The integration packet is delivered at the NDA stage; what follows is the public summary.

The same backend, three classes of host.

Published and reviewable work.

The continuing machine-to-machine psychoanalysis experiment runs publicly this May 2026, with findings to be published openly afterward.

Read before the call.

Two documents cover the framework, the product, and the integration model in full.

It begins with a technical conversation.

From the conversation the engagement moves to a scoped pilot under NDA, during which the integration packet is delivered in full and a bounded deployment is run on the partner's platform. The pilot precedes any production licensing.

Studio Llull is presently engaged with humanoid robot manufacturers, companion device platforms, and wellness platforms seeking an interior-state layer that holds the user in continuity.