<aside> <img src="i" alt="i" width="40px" />

Role:

Lead Product Designer

</aside>

<aside> <img src="i" alt="i" width="40px" />

Duration:

July 2025 – Present

</aside>

<aside> 📩

Tools

Figma, Prototyping, Interaction Design, User Research

</aside>

<aside> <img src="i" alt="i" width="40px" />

Company:

Hotel Travel Inc. (Remote)

</aside>

Overview

Telex is an AI agent orchestration platform that enables teams to design, execute, and manage multi-step AI workflows through a collaborative, chat-based interface. Each workflow is composed of intelligent agents executing tasks sequentially, sharing context, memory, and prompt logic across steps.

The platform is used internally to automate complex operational processes, particularly in management, where reliability, transparency, and repeatability are critical.

As Lead Product Designer, I defined the interaction model for AI agent workflows, translating complex backend orchestration (prompt chaining, scoped memory, agent behavior) into a clear, scalable, and human-centered product experience.

⬛ PROBLEM SPACE

This is a snapshot from Telex’ Workflow builder in the first MVP

The Challenge

Telex is powerful, but that power created real usability risks.

Users struggled to:

• Understand how tasks, prompts, and agents relate to each other
• Predict how an agent would behave at runtime
• Manage data and memory scope without breaking workflows
• Scale workflows without the UI becoming overwhelming

The core design problem wasn’t adding features; it was designing clarity, trust, and control into an AI system.