Retool

This page is under construction

Retool Agents are autonomous AI “workers” you build in Retool that use LLM reasoning to take actions — they call your APIs/queries/workflows, make decisions (or ask humans), and execute tasks end-to-end.

• retool.com/agents

I led the design of the product from early exploration to launch, working closely with engineers and product managers to define the user flows, interaction models, and visual design.

I’ve worked on various prototypes exploring different IDE concepts for building and debugging Retool apps. Some of these experiments have influenced the final product, while others remain as explorations.

Visualizing logic

Retool apps can get complex quickly with many interconnected components and stateful logic spread across multiple layers.

In an effort to help users understand and debug their apps, I worked on a prototype for visualizing data and control flow in a Retool app. The tree shows a graph of relationships between queries, variables and components on the canvas, allowing users to trace how data moves through the app.

The goal was to have a lightweight layer of visual feedback that is “always there” and helps users reason about app behavior as they build.

I joined Retool in 2021 to help build a completely new product — Retool Workflows. The team was just 2 people — me and a full-stack engineer. I drove product and interaction design, iterated on the core flow primitives (API blocks, logic, control flow, etc) and built most of the final Workflows UI in React.

• retool.com/workflows

Since then the team has grown significantly and Workflows has become a key part of the Retool platform. The point was always the same: move quickly, get honest feedback, and build automations engineers actually want to use.

Early on I prototyped several control-flow and transformer blocks in the Godot game engine because its canvas/graph APIs made it fast to experiment with runtime behavior.