When the Solution Gets Ahead of the Problem

You’re on a discovery call with a new client when the conversation shifts toward AI.

The meeting starts the way most do. You walk through their platform, their users, and where people tend to get stuck in the current workflow.

Some steps feel repetitive. Some tasks are still too manual. Support requests keep circling the same questions.

As you dig a little deeper, a few constraints start showing up in the conversation too.

They don’t want to add complexity for users. They don’t want anything that feels confusing or “too technical.” At the same time, they want the experience to feel faster. More efficient. Less manual.

Both ideas sit in the same sentence, like they naturally belong together.

Then the client brings up AI.

Not as something they’re exploring.

More as something they already expect to include.

They ask about your experience with it.

You pause for a second.

You explain that you’ve used tools like ChatGPT and Claude in your own workflow. Research. Thinking through ideas. Organizing information. Moving through design decisions faster.

But you’re also honest that using AI in your process is different from designing it into a product.

That part is still something you’re actively trying to understand.

The conversation continues, but the expectation is already there.

AI is going to be part of the solution.

The only question is where it fits.

Before the call ends, you try to get more specific.

You start pulling at the workflow a little differently — trying to separate what’s actually slow from what just feels slow in context.

Some steps come up again. A few manual handoffs. Repeated questions that support keeps answering.

But the answers still stay broad. Nothing fully mapped. Nothing clearly isolated as “this is the problem we need to fix.”

Just patterns.

Not clarity.

And that’s where the AI conversation starts to feel premature in a different way.

You leave the call with more questions than clarity.

Not just about whether AI can be used.

But about what you’re actually being asked to design.

That question follows you into your usual networking meetup.

It’s a mix of developers, UX designers, and people working across different product teams. A space where conversations usually turn into shared problem-solving more than formal explanations.

You bring up the discovery call.

The client’s expectations. The push toward AI. The lack of clarity around what that actually means inside the product.

As soon as you finish, the conversation opens up.

One developer talks about AI agents being used to handle multi-step workflows — not just responding to input, but completing a sequence of actions across systems.

Someone else adds that automation is usually more rigid — predefined steps that run when something triggers them.

A designer in the group simplifies it:

“Automation is predictable. It follows rules.
An agent is more flexible — it’s trying to achieve a goal, not just follow a step.”

That lands differently.

Because it highlights something you’re still trying to fully understand.

These aren’t just technical differences.

They change what the user thinks is happening inside the system.

The conversation keeps going, but now it shifts from definitions to implications.

If a system is making decisions, how does the user understand what it’s doing?
If something is automated, where does the user stay in control?
And if the output is unclear or unexpected, how does the user recover without losing trust in the system?

As people start sharing real examples from their work, the conversation gets more grounded.

AI suggesting incorrect outputs in workflows. Users not realizing something was automated until it broke something downstream. Features rolled back because trust dropped after automation was introduced too quickly.

And suddenly, the original client conversation feels different.

Because “efficiency” is no longer a simple goal.

Speed only helps if the user still understands what’s happening.

Automation only helps if the system remains predictable enough to trust.

And AI becomes risky the moment it replaces clarity with interpretation.

You sit with that for a moment.

Because it reframes the entire discovery call.

Not whether AI can be added.

But whether the problem is understood clearly enough to justify it in the first place.

Know a designer who feels stuck in UX? Feel free to share my site.

Helping UX Designers bridge gaps and grow