AI is going to build a lot of crap.

First off, I'm AI optimistic. Really, that headline be damned. It's true though. Look at the many of announcements in the last few weeks and months.

1. Anthropic artifacts: User - "I want a thing." AI - "Okay, here's your thing."
2. Figma Make Designs: User - "I want a thing." AI - "Okay, here's your thing, but it's really pretty too"
3. Replit Agent: User - "I want a thing deployed so anyone can use it." AI - "Okay, here is that thing, publically accessible - Everything is fine....."

See the pattern? Say what you will about me and all the other UX / Design Thinking leaders of the past decades, but this is literally what we've tried to not do.

UX is a practice of validation.

There is a lot to validating ideas, specifically strategy and research. Aspects that traditionally have taken as much, or more time and cost to gather data on, than visual design and development.

So why aren't companies thinking about using AI to lower the time and cost? First off, because it's harder. It's the human aspect of reasoning that AI currently struggles with. Second, it's not as cool as remaking a game like Snake or rendering photorealistic pictures.

AI lacks validation

This is a generalization, but an experience informed one, which is more than can be said about a lot of tech. It doesn't have to be that way. Instead of giving your team access to a chatbot, build a tool that aligns with a team's process.

Here's an example. Say you are an HR exec and you want to know insights on managers who inheirit direct reports after a layoff. Specifically you want to know if people who lost their managers are supported. How would you do that? Would you talk to the new managers, their new reports? Would you look at data and meetings, calendars and feedback? How long would that take.

Build AI tools

Instead of wishing you had a perfect answer, work with AI to find one. The example above is perfect for AI because it includes a lot of unstructured data. It's your experience in knowing what to look for that makes AI valuable. So instead of spending your time and money wondering, build a tool and roll that out.

www.chatagency.ai

Chat Agency AI is exactly that. I built this after Cisco restructured and laid off design leadership in April of 2024. The goal is to take the 'thing' you want and do the steps to validate it in an accelerated fashion. I like to tell people that if you have the idea in the shower, it can be validated by the time you finish breakfast.

It's new, it's not finished - or even polished - but the feedback has been positive and I hope it's turning bad ideas into great ones.

Give it a try and let me know.

--chad