Persistent memory
Research direction
Retention, retrieval, revision and deletion
We study how to retain, retrieve, update and delete information from conversations and tasks, including source records, temporal relationships and conflicting information.
Software practice: Agent-Me architectureAI Twins & personalization
Research & software practice
Managing identity, preferences and knowledge over time
We study continued use of personal agents built on knowledge, preferences and decision records. Agent-Me is the software implementation for this work. An AI Twin is managed by its user, not a digital copy or a replacement for their judgment.
Explore the Agent-Me implementationEmbodied interfaces
Exploratory research
Wearable sensing and robotic interaction
We study wearable sensing, human-machine interfaces and robotic interaction. This work is exploratory; no hardware has been released and robotic capabilities remain to be validated.
Digital cultural memory
Experimental R&D
Sources, dates and context in cultural records
Through CyberMemoir, we organize cultural records while keeping original sources, temporal clues and interpretations distinct, with gaps and review status recorded.
Explore CyberMemoir open-source workAreas of focus in memory systems
Four parts of a memory system: receive information, retain it, update relationships and use it in later tasks.
Interaction
Identify the question, the information available with permission and what needs checking.
Memory
Retain information relevant to the task, together with its source and date.
Structure
Connect related information, handle conflicts and support retrieval, revision and deletion.
Collaboration
Use retained information in later tasks, with user approval for actions that require permission.
Have a related research or engineering question?
Contact & conversation