MuleSoft: ‘We are Living Through the Most Important Transformation Humanity Has Ever Undergone in Its History’

The integration and connection between applications, services, databases and devices has become the cornerstone for organisations to be able to offer the modern and efficient services that we users – both professional and consumer – enjoy on a daily basis.

These processes have been MuleSoft‘s speciality since its founding in 2006, an organisation that was acquired in 2018 by Salesforce. Since then, its technologies have been implemented and extended in virtually every service of the world’s largest SaaS provider.

The emergence of generative AI and agents has highlighted how important these integrations and connections are, as well as making good on the $6.5bn investment Salesforce made when it bought MuleSoft.

Today, the two organisations work very closely together, but MuleSoft also continues to operate independently with its own customers, as Ricardo Usaola, MuleSoft’s regional vice president for Spain and Portugal, discusses in the following interview.

The dizzying pace of innovation in the technology sector is once again evident when we compare this conversation with the one we had with Usaola less than three years ago, especially after the advent of artificial intelligence and agents.

-A few days ago you held your MuleSoft Connect AI Madrid event. What were the main messages you delivered to the audience?

The whole event revolved around agents. We are living through the most important transformation that humanity has undergone in its history, especially because of the speed at which things are happening. The changes that are taking place today are exponentially faster than those experienced at other times, such as the advent of electricity or, more recently, the adoption of the internet.

During my presentation I proposed several examples that seem like science fiction but are very real today. From swarm intelligence with ‘humanoids’ in manufacturing plants to robotaxis, we are seeing advances that were unthinkable just a short time ago.

From Salesforce’s point of view, artificial intelligence is impacting businesses as never before through agents. For example, human resources departments are going to continue to care about the workers, the workforce. But we are also going to have the ‘non-human resources’, i.e. the non-human resources departments. These are the technology areas that are going to have to manage the entire digital workforce that works in the company: the agents.

Agents are going to communicate with customers and with the employees themselves, so someone needs to decide how they have to communicate with all of them. And we are already starting to talk about a very mature model in which it is also necessary to define how agents are going to communicate with other agents.

We have moved from agents that simply provide information to agents that act, for example by updating a database, to agents that are able to monitor other agents based on human-designed parameters of ethics or compliance. That is the business world where you will have to reconcile a human workforce with a digital workforce.

-As you say, this is truly advanced technology, but are we humans ready for it? Are we embracing these changes at the cultural and workplace levels?

Human beings have to adapt to these changes, it is something that is on the table. I think it is a reality that is going to come, whether we like it or not. Technologically, it is already happening. We already have technology that is capable of creating these agents and making them interact. It is true that a series of debates are inaugurated. If what you used to do with twenty people, now you do with five, what about the other fifteen?

‘All companies are having trouble scaling services to the level that customers are demanding‘

You can look at it from a negative point of view, as a simple cost reduction, but also positive: you have fifteen people who can do other work that brings much more value, that brings innovation, that can refine the way those agents are behaving, and so on.

The case of code development is a clear example of this. We, for example, have not hired more programmers because much of the code is already generated by the AI, but we still need developers to supervise all the code that is being generated automatically. And they are learning to provide the value that an AI is not capable of, to debug and to create new functionalities. AI helps to reduce time and people are prioritising. The real intelligence or creativity resides in humans, while the ‘business-as-usual’ is done by agents.

There is also another message here that is very important to give: all companies are having trouble scaling services to the level that customers demand.

-So it’s not about cutting jobs, it’s about increasing the volume of services that the market demands, scaling them up and customising them…

That’s right. It’s a question of scaling and hyper-personalisation. Agents, for example, are able to understand customers well and improve their experience through objective data analysis. And they can do it much more quickly and efficiently than humans.

-On the other hand, there is the creation of new profiles, new specialisations that bring more value and that should compensate for certain jobs that are being lost. Have you identified them?

It is a very complex question because it can be explained from the top, at the country level, where governments have to make decisions on how to manage this change that is coming through new responsibilities. Then there are the companies, and this is where Salesforce is telling its vision of the future.

In the end it is the organisations themselves that know their business and needs, so they themselves will find and define the new profiles they will need depending on where they start to apply the agents, because the truth is that they can be applied in any department and gradually extend their use to others.

We address any area of the company, be it human resources, sales, logistics, customer service, but it has to be the company that determines where it starts, where the agents can provide more value, what ROI is obtained in each of the areas and, at a later stage, how they communicate with each other through the multi-agents. And this is where new profiles and specialisations will be generated. Managers will be needed who are able to design how agents from different departments will communicate with each other. Many managers will have to reinvent themselves to adapt to other needs that will bring more to the table.

-Something similar ocurrs with the figure of the CIO, who continues to evolve in order to respond to the specific needs of each business area, but also to work on a unified vision of all departments. To what extent does this evolution influence your conversations with companies?

That’s a very good question. It used to be that the person who served a telco knew about telco, and the person who served retail knew about retail. However, this is changing. Now, even if I deal with retail, retail includes human resources, logistics, etc. So the sales person has to be trained to understand the whole business. For Salesforce, this means that the business has multiplied, as it is not just dealing with one person, but anyone in the organisation.

The Salesforce sales person must have telco-specific business cases if they are talking to a telco, but they must also be prepared to deal with HR business cases. All sales people need to understand the role of agents in a company.

In addition, the CIO is gaining more importance at the management level. In most companies, the CIO already reports to the CEO, which underlines its importance in the company’s strategic plan. The CIO not only responds to the demands of the business, but also proposes solutions and empowers the business in various areas.

Salesforce’s role has also evolved to serve the CIO, as it used to talk directly to the business, which then reported to the CIO. Now, everything is interlinked. The architecture of the applications is no longer niche, but requires an agent to talk to various areas such as service, procurement, logistics, sales, and so on. The CIO is the one who can interrelate all this. Therefore, CIOs will have a much more relevant and exponential role with the introduction of these agents.

-In an organisation such as MuleSoft, which is part of Salesforce but operates independently, you have to combine both factors. On the one hand, developing technology together and, on the other hand, putting your technology at the service of customers. How do you deal with this dilemma?

You have perfectly separated the two aspects because they are part of our go-to-market. An important focus for MuleSoft is the replacement of legacy technology. The hyper-automation and composability of businesses are now much more relevant than in the past. It is no longer just people who need access to specific data, but agents who need access to various applications.

When you have a single platform capable of handling APIs, integrations, connectivity, workflows, RPA, IDP, etc., you can really talk about hyper-automation. It’s not about having ten tools that have to connect to each other and when one changes, you have to change twenty things. It’s about having a single platform that can grow, scale and be agile.

Once you have that platform, you can move to the next level of maturity. If you don’t have the data intertwined, how are you going to launch agents? According to a Deloitte study, 95% of enterprise IT leaders have deployed or plan to deploy agents. Of that 95%, 94% say their problem is the lack of interconnection between applications and data. They have, on average, only 32% of applications integrated.

Point-to-point integrations don’t scale because when one piece changes, everything else has to change. MuleSoft allows companies to have a reusable architecture, like Lego pieces that can be reused. Within our marketplace and catalogue, we can include agents and manage them from a single point with MuleSoft. In addition, we can manage LLMs as they emerge in companies.

-This is where the sensitive issue of data privacy comes into play. We are no longer talking about sharing with the ‘outside world’ without consent, but between departments, so you have to carefully manage what information can and cannot be shared, is that right?

That is correct. Initially, companies use one or two LLMs to manage knowledge and deliver information. In six months or a year, there will be thousands of LLMs dedicated to specific topics such as human resources, banking, retail, etc. How do we bring the value of these LLMs into the enterprise, do we give them access to all enterprise information?

Our answer lies in the combination of Integration Fabric and Data Cloud to manage, anonymise and federate data. MuleSoft, on the other hand, facilitates access to external or Salesforce data and establishes barriers so that the information entering or leaving the Data Cloud is perfectly managed, governed and secured.

Within MuleSoft, we have built in a way to manage these LLMs, whether internal or external. An example of an activity that MuleSoft can perform is the management of LLM tokens. For example, Repsol uses MuleSoft to manage their internal LLM tokens. They have different categories, such as Silver and Platinum, and depending on the category, a specific number of tokens are allowed to be used.

-One of the headaches for organisations is the consumption of these technologies: the more queries or prompts with tokens and databases, the higher the bill. How do you deal with this problem?

Indeed, one of the risks of LLMs is uncontrolled consumption. People can ask a lot of questions and, without realising it, consumption gets out of control due to an error or any other reason. MuleSoft helps to manage and control the consumption of these tokens. When the user connects with a virtual agent, the agent will know that I am a VIP customer and will be able to use more tokens to provide me with a much more personalised response. If, on the other hand, I am a person who has logged on to the web, I will receive a response based on the number of tokens, the call and the information allowed based on my level.

-In addition to legacy modernisation processes, you are also reaching out to projects that start from scratch. What can you tell me about this?

This is the most important part of our business. The integration of Salesforce with other platforms is our business-as-usual, what we have been doing for a long time. The agent part is moving fast, but companies don’t immediately ask for 500 agents. They start with one, evaluate the business case, then move on to the second, and so on. MuleSoft brings value by connecting with SAP, updating databases, stocks, etc. This process is gradual.

MuleSoft’s business is not based on immediate figures, but on providing a strategic platform for growth. Normally, this growth will happen in agents, but it is necessary to be prepared and lay the foundations. MuleSoft helps to lay these foundations to avoid integration problems and to allow scalability. Any agent you want to set up can be a ‘drag-and-drop’ and connect to the desired system.

The word ‘act’ is key to MuleSoft, as it allows Salesforce to act both inside and outside its ecosystem. Initially, agents provide information, but as maturity increases, they are asked to act and execute. MuleSoft facilitates this execution outside of Salesforce natively.

Then there is the governance part. Once you have multiple agents, you can govern them, see who is using them, who is not, and so on. This is the future of enterprises and what MuleSoft is working on.