AI Radar

AI Radar is a data-driven visualisation tracing several months of my work with AI. It reveals the hidden rhythms of collaboration within a growing ecology of AI agents: my attention shifting between them, bursts of intense activity, temporary agents appearing and disappearing, and a shared body of knowledge accumulating over time.

The radar's circular sweep marks the passage of time. As it advances, it reveals where my attention is directed, which agents are active and how work passes between us. The coloured fields gathering near the centre represent our shared transactive memory: knowledge distributed across thousands of files that agents create, read and update.

The animation documents a shift from using AI as an isolated tool towards working with a system of agents whose contributions accumulate over time. By highlighting the temporal dimension of AI, it shows human intention, machine agency, and shared memory beginning to work as one dynamic system.

Draft v0.4. The figures are provisional and will change as I refine the analysis.

How I made it

AI Radar uses more than 5 GB of locally stored activity logs generated by Claude Code, Cowork and Codex. These logs provide a detailed record of every conversation, file operation, tool call and interaction between agents, each with a precise timestamp.

I analysed the logs with Claude Fable 5 and Opus 4.8. By synchronising activity across the three systems, we reconstructed a detailed portrait of our collective actions. The models also helped me design and program the animation, exploring how activity, attention and memory might be expressed through time, movement and colour.

AI Radar is therefore more than a data visualisation: it is also a reflection of the multiplicity of AI. AI generated the activity, helped analyse its traces and participated in constructing its representation.