Leap AI

In 2024, building an AI automation still meant coordinating models, prompts, data and tools that had not been designed to work together. Leap had grown from that exact problem: its founders found that generating an AI output was not the hardest part; maintaining all the orchestration required to bring it into production was.

The opportunity was to turn that complexity into a no-code product a team could understand and operate. A person needed to discover what to automate, connect the tools they already used, build the logic step by step and see what happened in each run, without having to think like an infrastructure team.

I joined the founding team in February 2024, shortly after Leap announced a $1.4 million seed round led by Founders, Inc. The product was in a period of speed and definition: we had to redesign the platform while the team continued learning what early customers needed.

I organized the experience around the complete lifecycle of an automation: discover or create a workflow, connect the necessary applications, edit its logic, run it and understand the result. This made Workflows, Integrations and the editor feel like parts of one system instead of separate tools.

I worked within a small team with a Formula 1 dynamic: short cycles, direct collaboration with founders and engineering, and decisions moving quickly into production. In addition to the core flows, I designed AI tools and a landing page that could explain the product's value outside the application.

To keep that speed from fragmenting the interface, I built a token-based design system. Components, icons, charts, illustrations and motion shared one foundation, making it possible to extend the product without redesigning every new capability from scratch.

The entry point to automations. I designed the experience for discovering, creating and managing workflows without forcing people to start from an empty canvas every time. Templates and use cases helped turn an intent —such as repurposing content or processing information— into a flow that could be adapted.

The interface also needed to communicate the operating state of each workflow: how it was triggered, when it had run and what needed attention. The goal was to make building and operating an automation part of the same experience.

Integrations were the bridge between AI and real work. I organized a surface for exploring tools, connecting them and understanding the capabilities each one brought into a workflow, reducing the distance between selecting an application and using it in the product.

The challenge did not end with a successful connection. The experience needed to cover permissions, credentials, states and configuration errors clearly, because an automation is only dependable when its dependencies are visible too.

The visual editor was the core of the product: a space for turning an intent into a sequence of triggers, actions, data and AI-assisted decisions. I worked on the canvas hierarchy, each step's configuration and the relationships between inputs and outputs.

I designed the flow for iteration: adding and reordering steps, configuring models or integrations, testing the workflow and identifying where it had failed. The goal was not to hide the logic, but to make it readable to someone who does not write code.

I built a token-based system to support the platform's rapid evolution. The foundation brought together color, typography, spacing and states with a library of components, icons and patterns for configuration-heavy surfaces.

The system extended beyond controls: charts, illustrations, motion and AI tools shared the same language. This allowed product and marketing to remain recognizably Leap even as the proposition changed.

The redesign connected the automation platform's central pieces: Workflows, Integrations, the Workflow Editor and the Design System that kept them coherent. I also extended that language to AI tools and Leap's landing page.

This case captures a specific stage of the company: when Leap was a no-code platform for building and running AI workflows. My contribution was turning technically complex infrastructure into an experience people could navigate, configure and extend quickly.