Case study 02 / Conversational product system

Designing continuity for 4,300+ AI companion users

How localisation, character design, persistent relationship state, privacy, and monetisation were combined into a repeat-use Hungarian conversational AI product.

ProductBarátnőim
FocusConversational product system
StatusLive
RoleFounder · Product · Engineering
Problem → decision → evidence

For conversational products, model quality is only one layer. Continuity, identity, progression, and trust determine whether an exchange feels like an ongoing product experience.

A conversation is not a relationship

A companion product becomes interchangeable when characters sound alike, forget earlier context, and wait passively for the next prompt. The product needed continuity and progression without turning the interface into a visible collection of AI controls.

Make the relationship the product state

Barátnőim was built around distinctive Hungarian personalities and persistent context. A Love Meter makes progression legible, characters can initiate interaction, and visual content extends the experience beyond text. The surrounding privacy and coin systems make that relationship loop operable as a consumer service.

  • 01Design original Hungarian personalities rather than translate generic archetypes
  • 02Persist conversational and relationship context across sessions
  • 03Expose progression through a simple Love Meter instead of technical memory controls
  • 04Support character-initiated messages and visual content
  • 05Integrate encrypted conversations, private access, and Stripe-backed coin payments

A localised product with repeat-use mechanics

The live platform now serves more than 4,300 registered users and offers over 80 AI characters across multiple categories. Its value comes from the combined system: localisation, memory, progression, content, privacy, and continuous availability.

4,300+Registered users
80+AI characters
24 / 7Availability
Next case studyOktelo CV evaluation