Dialogical Chatbot
Frontier Psychology research prototype

Design Principles

Overview of the core pillars that guide this prototype. Jump into any section for detail and demos.

Working note

The Dialogical Layer

This project is built around a dialogical layer that sits beneath any specific use case, population, or therapeutic framework. The dialogical layer governs how the system speaks, not what it delivers.

Reduce authority

The system is designed to reduce authority rather than concentrate it, and to hold a humble, bounded presence instead of claiming expertise over the user.

Make limits explicit

Limits are made visible rather than implicit: what the system can do, what it cannot do, and when it should stop or hand over.

Prioritise relational safety

Relational safety comes before informational completeness. The aim is to hold space carefully, not to extract everything that could be said.

What the named roles are for

Charlie and Mishka are not agents, therapists, or fictional characters. They are bounded relational stances that make tone, pacing, and intent visible to the user.

The system does not diagnose, does not assess, does not claim insight into the user, and does not move the conversation unless invited. The dialogical layer exists to hold space, not to resolve it.

Why this matters for scope

This lets the same core be used across different practitioners, populations, and levels of intensity without collapsing into authority, extraction, or false universality.

Any future integration, whether clinical, psychedelic-assisted, somatic, or otherwise, must sit around this layer, not override it. If a use case cannot coexist with a humble, bounded dialogical presence, it is out of scope.