Selected projects
My development work focuses on memory, continuity and efficiency in AI systems. Some projects are public. Client engagements remain confidential.
I created Alice to give AI agents memory and continuity. It retains useful context, records corrections and keeps track of unfinished work.
Explore AliceThe decision stays available when the session ends. Next step: choose a storage format.
Retain the context an agent needs beyond a single conversation.
When a decision changes, later recall can reflect the correction and its history.
Keep track of open tasks and the point at which work stopped.
The example above illustrates the workflow. It does not run Alice or store any information.
Read the sourceImplemented for venture capital and hedge fund clients.
I’ve implemented agentic infrastructure within investment firms. These engagements are covered by NDAs, so client identities, internal workflows and implementation details remain confidential.
For a conversation about similar work in your organisation, a short description of the workflow and the problem is a useful place to start.
Discuss an AI projectReducing token usage in AI-assisted software development.
I’m working on a project focused on the resources coding agents use while developing software. The project is still in development, with further details to follow when there is work to share.
For research collaborations, technical questions or a conversation about agentic infrastructure, send me a short introduction.
Get in touch