Product Design + Dev
/2026
Agent Story
“Making AI visible”
AI agents do a lot — but most of it is invisible. Agent Story creates a clear record of what happened and why, so people can trust, debug, and get excited about what AI can do for them.
Role
Solo — Design & Development
Status
In Development
The challenge
AI agents do a lot — but most of it is invisible. Without a clear record of what happened and why, it’s hard for people to trust, debug, or even get excited about what AI can do for them.
The approach
Most tools that expose what an AI agent did look like server logs — accurate, exhaustive, and unreadable to anyone who isn’t already debugging. Agent Story starts from the opposite end: what would it look like if an agent’s work read like a story instead of a stack trace? The core move is separating intent from action. For every step, the interface answers two questions in plain language — what the agent was trying to do, and what it actually did — before exposing the raw detail underneath. That lets one view serve two audiences: a non-technical user skims the narrative and trusts the outcome, while a developer drills into any step to debug it. Everything is organized as a timeline with clear visual weight — decisions and tool calls get prominence, routine steps recede — so the shape of what happened is legible at a glance. The goal was for someone to scroll an agent run and feel what it did, not decode it.
The outcome
Agent Story is in active development. The timeline view and intent/action model are working against real agent runs; the current work is keeping long, branching runs readable as they scale, and building the affordances that let a developer jump from “this looks wrong” to the exact step that caused it. Next up is integrating with live agent frameworks so the story writes itself as the agent runs.