Asking the wrong question costs more than buying the wrong software.

No One Wants a New CRM

No one wakes up thinking, "I hope I get to migrate to a new CRM this year." Nobody wants another login, another implementation, or another six-month rollout.

What people want is simple: fewer mistakes, quicker answers, happier customers, less repetitive work, and a little more time back in their day.

Companies turn to software for the same reason they hire new team members or overhaul a process: to make work feel lighter. The CRM, the ERP, the platform — these are means to an end, not the destination.

Picture a sales team losing twenty minutes after every customer meeting, just shuffling data from one system to another. One solution is to replace the CRM. Another might be connecting the existing systems. Or perhaps the duplicate step shouldn't exist. The right technology decision only becomes obvious once you understand the work.

If you're in real estate, you don't have to imagine it. It's copying the same listing details into Newhome, Homegate, and your own website, then putting them together again for a high-quality exposé — four versions of one apartment.
It's chasing a buyer's financing documents through an email conversation because nobody has a shared view of what's already been sent.
It's re-entering the official details for the third time because the CRM, the valuation tool, and the contract template work in isolation.

The problem is that software is tangible. Work isn't.
You can compare features. You can request demos. You can calculate licence costs.
Understanding how work actually happens is slower, messier, and much harder.
So we often skip it.

For years, the software industry has asked, "What system do you need?" That's the wrong question. The real question is: Which part of the work is tedious and time-consuming?

Static Software Is Becoming Less Valuable

Ten years ago, software vendors competed by saying: "Here's how our tool works." What they didn't point out is that you'd need to adjust your business to fit it.

That way of building software is becoming less relevant. Not because companies are fussier, but because AI is changing what's possible. Software used to depend on fixed rules. AI, on the other hand, is increasingly able to work with context rather than simply following fixed rules. That allows software to adapt more naturally to how a business already operates.

This changes the main competitive question.
It used to be:
Who has the best workflow?
Increasingly, the question becomes:
Who understands your workflow best?

That's a fundamentally different way of competing, and most of the industry is still playing catch-up.

Almost Every AI Project Starts Backwards

This is where most companies stumble. They leap to "let's use AI" before ever asking, "where are we actually losing time?"

MIT's NANDA initiative recently analysed more than 300 public AI deployments. Their conclusion was surprisingly simple: the biggest challenge wasn't the technology. It was the inability to adapt workflows, retain feedback, and improve over time. (MIT, The GenAI Divide: State of AI in Business, 2025)

In other words, the tools aren't the real bottleneck — the sequence is. Companies bolt AI onto a workflow that was never properly mapped out, and then wonder why nothing works right.

Technology must adapt to the work, not the other way around.

Process Comes First

Even the best software won’t do the work for you. It stores information. But someone still has to decide what happens next — which email goes out, what a document should say, who needs to be looped in. Those decisions still fall to people.

That's exactly what makes AI interesting.
Not because it replaces software.
Because, for the first time, it starts participating in the work itself.
Ironically, AI doesn't reduce the relevance of understanding work. It makes it more important than ever.

That's why I now think differently about software projects. Instead of starting with technology or product, I start with the work. Understand it first. Simplify what can be simplified. Then decide whether the answer is software, an integration, AI, or nothing at all.

To get started, I ask a handful of simple diagnostic questions:

• Where are team members losing the most time each week?
• Which tasks create the most frustration or errors?
• What information has to be copied or re-entered the most?
• Where do things regularly slip through the cracks?
• What is working well?

These questions help uncover the real bottlenecks, so any technology decision is anchored in how people actually work.

Final Thought

The tools are out there. The technology exists. What's missing isn't capability—it's the discipline to ask what should get easier before reaching for the next thing to buy.

The future won't be AI-first. It will be work-first.

Written by Fabio Mosimann — co-founder of Gridwork and Brixel. I have spent enough time around software projects to become convinced that they shouldn't always start with software.

Thinking about what should become easier before deciding what to buy?
Tell us what's slowing your team down.