AI will not save the legal industry. It will expose who never had a strategy

There are moments when it feels as if everything revolves around Generative AI. It has become the theme of events, the favorite sales pitch, the recurring subject in leadership meetings. Yet the more I talk to partners, directors and managers, the more I realize we are caught in a kind of self-deception. AI has turned into a symbol of movement instead of transformation. The problem is not in the technology. It is in our rush to look modern without changing anything in the structure that actually sustains legal work.

Today AI is treated like an accessory. A shiny decoration placed over operations still full of bottlenecks, overlapping tasks, manual decisions and management based on intuition. Many are excited about copilots while their firm’s data is still spread across spreadsheets, emails and disconnected systems. It is like trying to install computer vision in an environment that has not yet learned how to see itself.

The truth is that most are using AI to automate old tasks rather than create new possibilities. They continue operating with the logic of doing the same but faster. AI is not about speed. It is about changing the way we think, decide and deliver value. That demands more than technology. It demands culture, methods and organizational design. Without these elements, AI will only accelerate the delivery of something that is already outdated.

A colleague told me that AI helped him finish in forty minutes a task that normally takes four hours. Impressive. But no one asked if that task even needed to exist anymore. That is where waste hides. We are optimizing what should be eliminated. Automating what should be redesigned. Perfecting what is already obsolete.

This is the blind spot of technology adoption. We keep applying AI with an analog mindset. We expect it to confirm what we already know, just more efficiently. We want it to reassure our theses rather than confront us with better hypotheses. And when AI produces a creative or provocative suggestion, the instinct is to reject it, as if it were wrong simply because it challenges our repertoire.

I witnessed this in a complex corporate deal where an AI agent was asked to suggest risks and alternatives. At one point it presented a possibility that no one in the room had mentioned. Silence followed. Then discomfort. The first reaction was doubt. The second was a list of reasons why it was technically valid but unlikely. The third reaction came from one of the partners and changed the entire mood. He asked why none of us had thought of that first. At that moment the discussion shifted. It was no longer about AI being right. It was about expanding the team’s field of vision.

I have been in many rooms where the simple act of disagreeing creates tension. Lawyers are trained to defend with conviction, not to listen to what challenges their point of view. When that challenge comes from AI the effect is even stronger. It is not about functionality. It is about identity. AI in this context does not only confront a thesis. It questions the symbolic place of those used to having the final word. That is why it is as necessary as it is uncomfortable.

This is probably the main difference between those who will grow with AI and those who will remain stuck in pilots that never evolve. It comes down to the willingness to be challenged by it. AI only becomes strategic when it forces us to think better, not when it only helps us move faster.

There is one certainty. You do not need AI if you do not know where you want to go. No tool fixes the absence of vision. What I see today are firms trying to digitize a mindset that no longer works. They want to look innovative without letting go of decision structures that block real movement. They adopt AI like someone adopting a pet. They give it a name, show it off to clients, but never integrate it into daily workflows, never train the team to interact with it, never review processes, never allow themselves to be deconstructed.

Meanwhile clients evolve. The complexity of business evolves. The pace of disputes evolves. And the legal industry is still debating whether AI should review contracts. That is no longer the question. The real question is how AI will be used to do what has never been done before. And that begins in a place few are looking at, the foundations.

It makes no sense to invest in AI if you do not know where your data is. It makes no sense to have a legal copilot if governance still depends on politics rather than objective criteria. It makes no sense to talk about prompt engineering if partners still treat innovation as a cost center. And above all it makes no sense to place AI on top of a legal operation that has not learned to let go of what no longer serves it.

The firms that will thrive by 2030 will be those capable of integrating AI as an autonomous creative force rather than a productivity shortcut. This will only be possible with a true strategy. With redesigned processes. With a revised mindset. With rebuilt operational cores. With a new leadership pact.

You can adopt AI as if it were just another software purchase. Or you can embed it as the foundation of a new business model. The difference between these two paths is simple. The first keeps you in the game for a little longer. The second places you at the center of it.

Looking ahead to 2026 the picture is already clear. And here is a spoiler. Firms that continue treating AI as a side project will soon face clients who already use autonomous agents to audit legal opinions, monitor jurisprudence in real time and compare pricing between firms. Some legal departments are already integrating these agents directly into their internal platforms, reducing the need for traditional consulting. The gap between what is delivered and what is expected will not be subtle. There will be no room for institutional excuses. Only for results or the lack of them.

For those starting out or feeling lost in the noise the smartest question is not which AI to buy. The real question is what you are trying to build that AI can amplify. Because if you do not know the answer, no large language model will provide it for you.

AI may not give you the right answer. But it will certainly reveal who is still asking the wrong questions.