Are We Replacing Clicks With Prompts?

One thing has been bothering me lately.

For as long as I’ve worked with CRM systems, people have complained about clicks. Too many clicks, too many screens, too many forms, and too many places where hope quietly dies between Save or Save & Close.

I honestly can’t count the number of workshops where someone proudly showed that a process had gone from eight clicks to four. Everyone nodded. Someone said “much better”. A product owner may even have smiled, which in CRM projects is basically a standing ovation.

And everyone agreed: fewer clicks were progress.

Enter the AI agents. 🤖

With AI functionality becoming part of business applications, salespeople especially found a new shiny thing. Suddenly, users who had previously complained about having to fill in a few fields were happily writing prompts to research, update, or manipulate data in the system.

At first, it was simple:

“How many opportunities do I have?”
“What is my average deal size?”

Then it evolved:

“Show me the opportunities I own that are likely to slip this quarter, explain why, and suggest which three I should follow up on first.”

And suddenly everyone is excited about chatting with a computer.

Don’t get me wrong. If the result is that data in Dynamics is updated correctly, I’m all for it. But I do sometimes wonder how we don’t see the irony. Some of these tasks are now taking longer, requiring more typing, and leaving more room for interpretation than the old “click-heavy” process we spent years trying to improve.

A simple update used to be:

1. Search.
2. Open.
3. Update.
4. Save.
5. Done.

No fireworks. No keynote moment. Just a user getting something done.

The futuristic version sounds more impressive:

“Find John Smith from Contoso, update his mobile number, check for duplicates, verify his relationship to the account, and save the changes.”

But it also sounds like I’ve just assigned homework to a very eager intern.

And yes, the agent might do it perfectly. But was it faster? Was it easier? Or did we just replace a few clicks with a token-heavy prompt and a new type of uncertainty?

Sometimes I think we are so busy admiring what AI can do that we forget to ask whether it should do it at all.

Background Agents Are Different

To be clear, I’m not saying agents are useless. Far from it.

Some of the best agents probably won’t sit in a chat window waiting for users to write perfect instructions. They’ll work quietly in the background. They’ll monitor risks, summarize meetings, prepare account insights, spot duplicates, suggest next actions, and warn us before something important falls through the cracks.

That makes sense.

I don’t need to chat with an agent to know that a customer has gone silent for 45 days while an opportunity is still forecasted to close this month. I just need the system to notice it and show me when it matters.

Not every interaction should be a conversation; sometimes I don’t want to explain myself to software. I just want to press the button and get the job done.

The Mouse Was never the Problem

Graphical interfaces became popular for a reason. Humans are extremely good at scanning information.

Open a sales pipeline, and your brain starts working immediately. You spot the deal that has been sitting too long, the opportunity that suddenly changed value, the salesperson with no activity, and the forecast that smells suspiciously like fiction.

Nobody needs to explain it. Your brain automatically processes the information.

That’s why dashboards, views, charts, tables, and forms still matter. Not because they are old-fashioned, but because they match how people actually think and because sooner or later, someone still needs to verify what happened, understand the context, and check that the right thing ended up in the right place.

– An agent works best when you already know the question.
– A user interface helps you find the question.

Is the Future Chat Only?

Agents will absolutely become, and already are, a major part of business applications. I just don’t think the future is agents instead of user interfaces.

I think the future is agents inside user interfaces, with clear and action-based outcomes.

Agents that prepare.
Agents that suggest.
Agents that warn.
Agents that summarize.
Agents that quietly reduce the amount of searching, remembering, and manual work required.

And who knows. Maybe in a few years I’ll demonstrate how buttons work in a system, and people will think:

“Damn, that’s a pretty straightforward and fast way of navigating information.” 😆