Pia / Pleiadi.ai
Pleiadi.ai / Pia is an internal Nocta R&D project exploring conversational AI, digital presence and human-like interaction through an AI companion / digital human interface.
Pia is not a Nocta service and not a finished product. It is an advanced internal AI interaction R&D project, shown here as evidence of how the studio thinks about and works with applied AI.
/ Why it exists
The question we're exploring.
The most interesting open questions in applied AI today are not only about answers — they are about presence: how an AI interface behaves over time, how it remembers, how it expresses state, and how a person experiences talking to something that is not a chat bubble in a sidebar.
Pia is the project where Nocta works through those questions directly. The aim is not a packaged product — it is to build first-hand understanding of how conversational AI, memory and visual presence fit together, so that the studio's client work is grounded in real practice rather than only in documentation.
/ What works today
Current state.
- Conversational AI and interaction architecture has been designed
- Web / chat interface and AI logic direction exist
- Presence and emotional-state concepts have been tested
- Unreal-based visual presence and digital-human experiments have been explored
Pia is intentionally described in terms of what is working today, not what is planned. The project is not production-ready and is not offered as a service or a digital-human product.
/ Approach
How we are building it.
Pia combines several lines of work that usually sit in different products: language, memory, voice, emotional state and visual presence. The intent is to understand the seams between them, not to shortcut past them.
- LLM-based conversation logic
- Memory, persona and RAG concepts
- Voice direction
- Emotional-state logic
- Unreal-based visual presence experiments
/ What this proves
AI depth.
Pia shows that Nocta does not only integrate other people's AI. The studio builds, tests and lives with its own AI systems, which is what allows us to make grounded decisions when designing AI workflows, assistants and agents for clients.
/ Curious about applying this in your company?