Introduction

First of all, thank you Jorge for the invitation and to Stefano.

Overview of the Talk

And I'm going to talk about AI agents. Do you know about AI agents?

A lot of AI people here.

AI Agents and Their Role

This is a part of a conference that we had recently with a group of HR people that gathered in Madrid to have their international get together. And this is an association of HR outplacement companies.

and we talk about AI agents.

Revolutionizing Technology with AI

So, as Jorge said at the beginning, I always like to say that with AI, the deck of cards was dealt again. So now everybody has the cards to automate, to improve their processes, to make things better, to make things faster, to make things cheaper. And it's not only, and technology now is not only in the hands of the big companies or the big procurement, companies with big procurement capacity, but it's for everyone.

So this is very interesting, especially if you're funding a startup.

The Future of Organizations with AI

So, I'm going to talk about the future of the organizations with AI agents and this is the starting phrase. The real threat is not AI replacing people, it's organizations failing to redesign themselves to work with AI.

So this is all about, this is not about technology, I'm gonna talk a little bit about how we build agents, we build agents and digitize, but it's about organization, it's about, it's not tech. This is about the structure.

So I'm coming from a background of managing companies in different places in the world and serial entrepreneur and all this, and this gives me a little background on organization, on processes, on planning, and investing in a company, as you all know, that helps you to learn a lot because you're putting your risk

Understanding AI Agents

upfront so we learn on that on that way as well so real quick you all know this but it's just to set the ground AI agents is not automation not at all totally different automation so we have

several types of AIs.

Types of AI

The first one would be the narrow, which is actually performs specific tasks. It's like a software, standard software, coding input-output process in the middle.

And then we have a generative AI, which is the content that we all know.

And we have the agents, and the agents have specific roles and The use of the agents is to integrate them into a process, is to integrate them into a workflow of the business and make them productive.

Purpose of AI Agents

And it's not to replace the human stated at the beginning, it's to empower him, to make it stronger, to make it fast, to make him more productive. So it's not replacement at all.

So we develop agents and we have, you know, there are many ways of, you know, the prompting structure and the blueprints of the outputs and, you know, lots of stuff on how to build the agents.

The 5Cs Process for Developing Agents

We use this 5Cs process.

We have the character, which is basically the role. Character is not really, but we use that to be, you know, to have the five Cs together. So it's the role that it plays as a customer service agent, for example.

The context, which we talk about or Alejandro showed us, in the context side, so it's in the context of regulation, in the context of pharmaceutical, in which context it's working on and we have the content and the content I'm going to double click here in content because companies

have their information their procedures their standard operating procedures their compliance documents their financial information their strategy everything is written everything is in the form of documents and all those documents are the what is called the single source of truth for that specific company So the content is what really changes and the agent is designed to act upon that content and provide the answers required using that specific content.

So the content is really the base or the DNA of the companies.

Then we have control. So definitely the control is governance of the data, of the information, but the control of what the agent can say.

So Alejandro just explained how you need to have this are more night with everything that all around to protect and it's still it's very hard because still it can you know it's easy is easy still is it possible to drown down the data out of your all your specific knowledge that you don't want to be shared so if you work in a closed environment

with agents that are not in this type of public platform, but build the agents separate, not connected to internet. I have the agents working, interacting with the content, with the single source of truth.

a better layer of security, but still, control is very tough and sometimes it behaves funny. I mean, sometimes it's just unpredictable.

So, depending if you have a big plan of security, it can still have issues.

And we have connectivity and that connectivity is the integration with the real work, the real world. So the agent should be able, if has information, a quick example and let me just check the time.

I'm doing fine, six minutes. You can have in the knowledge base of the agent a lot of information, a single source of truth for a given company, but you may want to process data, you may want to process information, you may want to process contracts, you may want to process medical data, you may want to process results from tests, from blood tests, for example, and that data is very sensitive. So you don't want that data to go out anywhere or to be stored anywhere else. than in your system.

So for this specific, you have to have either bring the information into what we call a vault, a vault, secure vault where you have an AI with your data. So bring the information in, work with it, You upload it or you do it through an API.

So it's closed, and this is what we do, work in a closed environment. So it's the same, same challenges, but still, but in a closed environment.

So these are the five Cs we use for the agents, to create the agents.

Implementing AI Agents in Organizations

And then now we go to the, you know, the implementation. So, okay, nice.

AI agents can do a lot of things. Cheaper, faster, no mistakes.

Okay, but what, you know, how do I use them? How can I make use of them in my organization? Because everyone wants to use AI.

Challenges in Adopting AI

Are you using AI? Yes, everybody's using it. But really, it's not that the case.

And it's a matter of... but educating the customers in the market. There is still a lot of information that, you know, a lot of stuff to agree on before even starting using AI in organizations.

So for the implementations, so we have the first part, which is risk killing. So it's all organizations this afternoon.

We were in a previous meeting with Miguel earlier today and we were talking about notifications from the courts. And these are these people that are working on this specific, let's say, data extraction. environment for 25 years.

So it's like, OK, you know, what about the AI? Have you heard about it? So it's hard in that sense to move.

And that's changed from people that are being working the same, doing the same thing for a long time. But still for new people as well, it's a challenge because it's like, hey, are you going to replace me? Or am I going to have more work to do? Because yeah, you put AI, I can do more things, so I will work more.

It doesn't mean that I'm going to go home earlier. So, there are all these challenges on the implementation and the challenge of being trained, of knowing how to use it, knowing what not to put in. You don't want to put company data in open AI platforms. You don't want to put customer's data, you don't want to put your organization's data out there.

Hybrid Organizations and Training

So the colleagues are now, so the agents become colleagues, become, the organizations are becoming hybrid. So hybrid organization having human plus AI agents to make, you know, to make the work better.

It's a tool, it's like a calculator, it's still in Excel, you know, it's the same thing. But people need to know, to learn how to use them.

But they have to be trained. So all those five Cs, they have to be trained.

So how do I use this agent? What is this agent going to do?

And you know well about programming GPTs, for example, which is very common. But they need to be trained to be trained on what they need to do.

Strategic Decision-Making in AI Adoption

So this is the main part of the presentation and is that the decision of using AI is not a technical, You need to think, rethink the organization. So it's first structure and not technical first because if you take technical and tech guys like to, I mean I'm sort of a tech guy as well, so like to go on the technical decision and it's not that.

So it's organization. I was talking about processes and some experience I had in the past in different organizations and that gives you a different view of how to implement, how to build a hybrid organization using AI agents.

So I was saying that

HR are becoming more, are having more responsibilities with the technology, with the technical side. So HR in some organizations, and I just can't remember right now because I read this yesterday or the day before about HR manager or HR. people person in a big company in the US becoming managing also the technology.

And this is because it's moving towards organization more than technical.

Use Cases and Future Outlook

These are some of the uses for HR in screening and onboarding, training, mentoring, engagement predictions, retention. mentoring and policy compliance.

Policy compliance is a very good use case for an agent for any organization.

We know, we heard that the market is growing, it's ready to fly, to rocket high sky, but there are predictions on numbers from Bloomberg, Accenture. At the end of the day, which is okay, you have these predictions, okay, this is going high.

Adoption Challenges and Opportunities

I believe that it's here to stay, but the adoption is not as simple.

And we have to start, as I used to say, if you want to cover all this area, then start from this little corner here, and then start growing it. Because implementing it as trying to change from one day to the other is not simple, it's not easy, and it's not even recommended.

So I have here one slide, which I'm not going to read through it. but it's about the challenges of adopting AI, adopting AI agents in organizations.

And it passes from having the alignment, the decision to implement AI or AI agents should come from the top of the company. decision from general manager from top management has to be you know part of the strategy and that it flows down otherwise if there is no buy-in that doesn't doesn't really work so the strategy the process yes having the process ready yeah big companies have the process supposedly to be all lined up and written and organized but many times they are not updated

Data quality, governance and trust, and there are still gaps in the skills.

So for my minute 14,

Key Questions for Organizations

I'm going to put this one, which is basically questions to ask to the organizations to help them decide how to start, how to move. So just a few questions to help. drive the conversation about AI, adoption of AI, and how AI is changing the, you know, can change the landscape of a company.

I think we're going to share this presentation, so you will have this information.

Okay, so that was 15 and 20, not counting this.

Conclusion

One big applause.