Visualizing high-dimensional agent trajectories

An autonomous agent does not move in a straight line. It explores a high-dimensional state space, backtracking, forking, and looping. Visualizing this execution path is one of the hardest problems in AI UX.

Beyond the Chat Bubble

The standard "chat" interface flattens a complex tree into a linear list. This hides the agent's reasoning process. Did it consider alternative tools? Why did it reject path B?

The DAG Representation

We visualize agent execution as a Directed Acyclic Graph (DAG). Each node is a state (thought, tool call, observation). Each edge is a transition probability.

Embedding Space Projection

To help operators understand semantic drift, we project the agent's context window embeddings into 2D space using UMAP. This allows you to visually see when an agent is "drifting" away from the system prompt's intent.