Ask Me Anything
Working with the Machine Learning team, I designed the interface for a RAG-based knowledge tool running on gpt-3.5-turbo, which quickly became one of the most widely praised tools amongst customer service agents.
UX Design
Adam Yuras
Product Engineer / Designer
An AI Optimist and Hype Skeptic.
AI is transformative, but creativity remains a uniquely human ability.
Making performant, high-quality products still requires a fundamental understanding of design and engineering. AI can help us build quickly, but without a vision, you're just going nowhere fast.
What I've Done
I've never been afraid to learn something new.
My mix of design and engineering skills give me a wide-ranging perspective that few others can claim. At Comcast, my journey took me from senior UX designer on our chat and voice apps, to product engineer for our universal AI assistant.
UX DesignerProduct DesignerDesigner & EngineerProduct ManagerAI Product Engineer
What I've Built
A timeline of AI products at Comcast.
Auto-Enhance
To improve the quality of our chat users' messages, we integrated gpt-4o-mini to automatically "enhance" their grammer and spelling. It currently improves the quality of over 1 million messages per month.
Design & Prototyping
Explain My Bill
Before "Agentic" was a buzzword, we created a multi-step AI tool to fetch and calculate a bill comparison, and transform it into a clear, structured output to assist agents through billing discussions.
AI Engineering
Auto-Disposition
Adapting the Python-based Google Agent Development Kit (ADK), we created a tool which used gemini-2.5-flash-lite to automatically label resolution state and sales outcomes.
AI Engineering & Implementation
Mercury Assistant
The "Universal Assistant", designed to serve multiple platforms across chat, voice, and retail. Also built using ADK, it contains a suite of AI tools using gemini-2.5-flash and gemini-2.5-flash-lite to assist with billing, sales, and customer support.
Design & Engineering
My Perspective
A realistic view of AI.
I've been fascinated by AI tools ever since the early days of Dall-E and GPT-3. I've always known there was great potential in the technology. But years later, there remains only a few areas where it's really proven its usefulness. The truth is, without serious intervention, LLMs produce generic, uncreative results.
I've watched other teams fail because they over-estimated AI's critical and creative thinking abilities, and under-valued people at the same time. My goal is to help businesses and teams from getting stuck waiting for AI to deliver something it can't.
Get in touch