Do You Really Need AI?

Over the last few days, I’ve had the same conversation on repeat: how to implement AI in law firms, companies, and institutions.

It feels like AI has become the only option on the table.

And let me be clear: I love AI (and I’m also scared by it). I’ve worked with it long before it became a buzzword. I teach it in master’s programs. I’ve seen it perform tasks that were simply unthinkable just a few years ago. It can speed up processes, automate repetitive work, and unlock entirely new scenarios.

And this is just the beginning. In 6, 12, 18 months, platforms will be more powerful, more integrated, and more accessible. The competition will be fiercer. The pressure will grow. The stakes will rise.

So yes, the potential is real.

But.

There’s also a growing tendency I find troubling: the rush to implement AI just because it’s AI. Not because it solves a real problem. Not because it adds value. Just because it’s trending. Because everyone else seems to be doing it.

So before jumping into yet another AI initiative, I suggest asking yourself a few uncomfortable, but essential, questions.

1. Do you have good and reliable data?

A couple of days ago I was speaking with a friend who works in a large law firm. Associates are required to log exactly eight hours per day. Not seven. Not ten. Just eight. And weekend hours? Not allowed, even if people work through the weekend, which lawyers often do.

What ends up in the system doesn’t reflect reality. Maybe it’s off by 10%. Maybe by 30%. Either way, the data is unreliable.

Now imagine building an AI model to track productivity or inform staffing decisions based on that dataset. What do you get?

Garbage in, garbage out. You’re not optimizing reality. You’re optimizing fiction. The result? A mess. Possibly worse than what you started with.

2. Is AI really the most efficient solution?

Sometimes the real issue isn’t a lack of AI. It’s the presence of broken processes.

If your workflows are convoluted, your communication unclear, or your services over-engineered, AI won’t help. It will just automate the confusion.

Would redesigning your process get you to the same result faster and cheaper? Would better collaboration solve the problem more directly?

AI implementation requires alignment, training, infrastructure, and long-term commitment. Are you ready for that, or just looking for a quick fix?

3. Have you accounted for the real costs?

AI is not a plug-and-play tool. It’s an ecosystem.

Beyond license fees, you’ll face integration headaches, training sessions, change management issues, and ethical or legal risks. You’ll need time for testing and iteration. You might need new roles and competencies. You might be pushed to invest more in cybersecurity and IT.

Have you considered the full picture? Or are you underestimating the invisible costs?

4. How messy is your current process?

If your internal processes are unclear, undocumented, or full of exceptions (and in most cases they are), AI will not clean them up. It will amplify the mess.

You can’t automate chaos. You’ll just get faster chaos.

Start with a process map. Understand the flows. Identify friction points. Simplify where you can. Then, and only then, think about automation.

5. Are your people trained, and truly on board?

Have you invested in digital upskilling? Do your teams understand how LLMs work, what RAG means, and how to use these tools responsibly? More importantly: do they trust them?

Even the most advanced system will fail if the people using it are skeptical, confused, or simply overwhelmed.

By the way, training people takes time, commitment, and a real learning culture. But if your actual priority is asking associates billing 2000 hours a year, then training isn’t a priority - no matter what your website or social media posts say.

6. Is your governance structure ready and committed?

Successful AI implementation is not just an IT job. It’s a governance issue.

You need decision-makers. Champions. Time and budget. A clear vision. A willingness to iterate. Alignment among key-players.

If your managers, IT leads, and innovation officers are not on the same page, AI won’t stick. You’ll have a proof of concept that never scales. Or worse, an expensive tool nobody uses.

7. Will your platforms talk to each other?

AI tools often rely on data integration across systems. But what if your platforms don’t communicate? What if you need APIs that don’t exist? Or worse—custom integrations that cost more than the software itself?

Compatibility isn’t a detail. It’s the foundation. If you haven’t mapped your system architecture, you may be solving one problem while creating three more.

8. What if your AI vendor disappears in two years?

It’s a real risk. The AI market is full of startups, and many won’t make it.

What’s your backup plan if your vendor shuts down, pivots, or gets acquired?

Do you own your data? Can you export it easily? What’s your lock-in exposure?

The goal isn’t just to buy a good tool. It’s to build a resilient ecosystem.

9. Do you want to substitute people, making them more productive or something else?

This question often gets swept under the rug. But it’s fundamental. And it may change considering the role.

What’s your vision? Are you trying to cut headcount, or to free up time for higher-value work? Are you aiming for cost reduction, or for strategic transformation?

Your intentions shape everything: the tool you choose, the message you send, the impact on culture.

10. Are you ready to run the race?

AI is not a one-off decision. It’s a shift in mindset.

The moment you implement your first model, you’re entering a new game. A game that moves fast, and keeps getting faster.

That means continuous learning. Governance that evolves. A culture that tolerates failure and rewards experimentation.

And yes, it means becoming, at least in part, a tech company. A tech law firm. A tech-enabled service provider. Whether you like it or not.

I’ve seen the wonders of AI up close. And I’m still amazed by what’s possible.

But I’m also wary. AI is often treated like a shortcut, a status symbol, or a silver bullet.

It might be the right solution. But it might also not.

As managers, partners, consultants, and professionals, we have a responsibility to look beyond the hype. To choose what works, not what’s fashionable.

Sometimes, that means taking a bold leap forward.

Sometimes, it means taking a thoughtful step back.

The real challenge isn’t about being first.

It’s about being right.