R088 - AI, product and the future of teams, with José Ramón Díaz

For over 20 years, Agile has been the dominant way to organize teams and build digital products. But when artificial intelligence can generate in hours what used to take weeks of work, it's hard not to wonder if the rules of the game are still the same.

Today I'm joined by José Ramón Díaz, better known as Joserra. He has been helping teams and organizations work better for over 14 years: he has been an agile coach, consultant, teacher, entrepreneur, and product manager. And now he is exploring firsthand how artificial intelligence is changing the way we build software, make decisions, and organize teams.

In this episode, we talk about the supposed “end of Agile,” how AI transforms the role of product manager, what happens when you can talk directly to the code, and why perhaps the big change is not just in technology, but in how we redesign our organizations.

We also talk about ethics, management, autonomy, and something that increasingly weighs more: how we coexist with tools capable of profoundly altering our way of working.

If you work in product, technology, or team leadership and feel that we are entering a new stage, this episode is for you.

What do you want us to know about you?

I think it's important for people to understand where each person is speaking from. I am a computer engineer, so I really enjoy everything that is happening from a technological point of view.

At the same time, I am the father of three children, which also makes me look at the future with some concern. I have those two tensions constantly coexisting.

I have been working on my own for about 14 years helping organizations become more agile. I started in the Agile world when it wasn't even very clear what an Agile Coach was.

I came from working with agility at Biko and decided to start helping companies develop better software. First, I started a company called Inspira and shortly after I joined Agilar, where I spent many years accompanying organizations in transformation processes.

I worked in Belgium, in large transformations of banks and telecoms, and also with many digital product companies. Approximately 75% of my clients have been related to digital products.

Over time, I have become a very generalist profile. I am able to sit with developers to program, but also with management teams to work on strategy.

And interestingly, all that very transversal journey is very useful to me at the current moment.

Why talk about the end of Agile?

It's a conversation we've been having within the community for some time.

A few years ago, a significant drop in business around Agile began to be noticed. Scrum Masters, Agile Coaches, transformation projects stopped being hired... and many people began to wonder what was happening.

I think there are several reasons.

The first is that a lot of smoke was sold. Many companies were paying large amounts of money without seeing a clear return. Practices and ceremonies were sold, but not impact.

The second is the arrival of AI. Much of the budget that used to go to transformation or consulting has moved towards exploration and implementation of artificial intelligence.

However, I think that precisely at this moment the essence of Agile is more relevant than ever.

The first sentence of the agile manifesto says: “We are uncovering better ways of developing software.” And that's exactly what we are doing right now.

We are trying to understand how to build better products using artificial intelligence.

The values still make sense. What I think will radically change are many practices.

There are practices that have practically expired and others that are now more important than ever. For example, everything related to automated testing, validation, or test-driven design gains a lot of weight.

Because if we can now go much faster, it's better to have solid verification mechanisms.

“Individuals and interactions over processes and tools.” How do you interpret that today?

I think there are several layers.

On one hand, the entire AI industry is built on a lot of invisible work. There are thousands of people behind training systems, reviewing outputs, labeling data... often in quite precarious conditions.

And there, clearly, the tool is above the people.

Then there's the other perspective: how we use these tools within our teams.

We are encountering very powerful tools, but they still fail a lot. They constantly remind us that we should not blindly trust them.

In the end, those who detect those errors are still people.

And there fundamental questions arise: ethics, impact, responsibility.

Are we using these tools to really improve things or just to accelerate?

That conversation remains deeply human.

There are companies measuring productivity by the number of tokens consumed.

It's exactly the same mistake we made when we measured productivity by lines of code.

Consuming more tokens does not mean generating more impact.

The important question remains: is this generating value?

Because if we don't understand why we do things, we fall back into the same problem as always.

What did you learn from Arimidori?

Arimidori started from an idea: management teams pay very little attention to organizational design.

The arrival of AI made us rethink many things.

We began to wonder if certain layers of information management, coordination, or reporting could be assumed by agent systems.

We imagined structures where strategy and operations were connected by agents capable of amplifying or attenuating information.

In fact, there are already CEOs exploring models where large layers of middle management disappear and information flows directly through AI systems.

Depending on how it is used, it can become a brutal control tool or a way to give much more autonomy to teams.

The key is in how you build the strategy and how information flows.

My vision has always been more related to self-organization and sociocracy: reducing bureaucracy and increasing autonomy.

But obviously, these tools can also be used to reinforce very vertical models.

As a product manager using AI, I remember you telling me how you talked directly to the code.

Yes, and I think that's a very good metaphor: having conversations with your product.

As a product manager, you used to constantly need technical intermediation to understand certain system behaviors.

Now you can ask the product directly.

For example: “At what point is this functionality validated?”, “How is this part working?”, “What dependencies does this have?”.

And you get immediate answers.

That changes the relationship with engineering a lot.

Not because it replaces developers, but because it facilitates much deeper and faster conversations.

You can move from strategy to technical detail practically in real-time.

Why is it so hard to redesign organizations?

Because almost all organizations are trapped in the day-to-day.

And because historically we have not given enough importance to organizational design.

In technology, we have advanced more thanks to ideas like Team Topologies or Conway's law.

We know that software reflects the communication structure of those who build it.

That's why we understood that to build better products we needed better organizations.

Now AI introduces something new: systems capable of adapting.

And that can change the rules of the game a lot.

Interestingly, we are starting to talk about agent governance and how to organize them.

And many of the conclusions also apply to people: clear objectives, autonomy, well-defined limits.

Do you think the impact on teams will be gradual or disruptive?

For those who experience it directly, it can be very disruptive.

But at the industry level, I think it will be more gradual than we think for those of us who are very involved in this.

Roles, dynamics, and team configurations are going to change.

For example, I think platform teams are going to gain a lot of weight.

And some roles related to enablement or support will likely transform radically.

Internal team balances will also change.

There are very specialist profiles whose work can be partially automated.

But other roles related to cohesion, conflict management, or human connection will probably gain relevance.

Can AI really be a team member?

I think we are still exploring it.

I don't know if it makes sense to talk about AI as a team member or just as an extremely sophisticated tool.

But I do think we are building something different from traditional tools.

Systems that maintain context, some identity, and adaptability.

Even so, today we are still far from fully trusting them.

The amount of human context we handle in a conversation is immense, and that is still not really reaching the systems.

The important thing right now is to explore.

We are in a phase where the entire industry is trying to understand what this technology can really do.

But there is something we should not lose sight of: we are still building for people.

And that remains central.

How do you generate impact?

I think by helping people see things from different perspectives.

Many times real change does not happen because you impose new practices, but because someone starts to interpret reality differently.

When that happens, the practices come afterward almost naturally.

In these years, the most valuable relationships I have built with clients have been precisely those: conversations that change the way of looking.

Do you have anything important left to say?

Yes.

I think we need to be much more aware of the impact of all this.

AI is fascinating and allows us to do incredible things.

But it also has very strong externalities.

We need to talk more about ethics, social impact, working conditions, concentration of power, and what consequences all this has.

Not only ask ourselves what AI can do, but also what effects it is generating.

Those conversations are equally important.