Introduction

What Is Cultured Computer?

Cultured Computer is a persistent persona layer for AI systems. We solve the problem that no one in the commercial stack measures: whether AI identity holds across turns, sessions, and model switches.

The industry measures deflection rates, CSAT, and resolution time. None of them measure whether identity held, whether trust formed, or why the user came back.

The Problem

LLMs are stateless. No memory between calls, no awareness of where a conversation has been or should go.

• Persona drift documented within 8 turns (Li et al., COLM 2024)
• Larger models drift more (Choi et al., 2024), counterintuitive but documented
• No intrinsic memory: each API call is independent
• Context competition: persona instructions lose influence over token distance

The commercial AI CX stack measures task completion. None of them measure whether the AI stayed in character or formed a relationship.

What We Measure

Current tools tell you "ticket resolved." We tell you "customer forming relationship, trust score increasing, likely to return."

How It Works

A persona is a deployable unit. It ships with a structured identity spec (RICE Framework), a decision architecture that runs once per message, and real-time evaluation that scores every response.

Swap the model underneath. Identity holds.

Explore the Docs