AI-Powered Interaction
AI avatars and agent systems designed as experiences, not isolated technology.
I design interactive avatars for museum contexts and build agent-based systems where conversation, behavior and interface need to work as one coherent audience journey.
AI avatars for cultural and visitor experiences
An AI avatar in a museum must do more than answer questions. It needs an understandable role, an appropriate tone, boundaries, recovery behavior and an interaction flow that works for visitors with different levels of knowledge and confidence. My experience includes creating AI avatars for museums, connecting conversational behavior with the surrounding interactive experience.
Agent-based systems
I can design and assemble systems in which specialized agents coordinate tasks or behaviors. The work begins by defining responsibility, inputs, outputs and escalation paths for each agent, then shaping the audience-facing layer so the underlying system remains legible and controllable.
- Avatar role, personality, boundaries and conversation flow.
- Agent responsibilities, orchestration and handoff logic.
- Prompt and behavior prototyping around a defined use case.
- UX/UI for conversational and agent-assisted experiences.
- Testing of failure states, unclear requests and recovery paths.
Experience-led implementation
My background in game systems, real-time development and UX/UI informs this work. It helps treat the AI layer as a responsive system with state, feedback and visitor expectations, rather than as a detached chat window.
Responsible scope
Every implementation should make capabilities and limitations clear. Agent autonomy, stored information, moderation, privacy and human escalation need to be defined according to the context before public deployment.