Appian: ‘AI is no Longer a Toy but Enterprise-Grade Quality’

Since Appian officially landed in the Spanish market, this software manufacturer has experienced continuous growth in terms of both business and capabilities and functionalities.

Its arrival on the market has not changed significantly since then, although its philosophy and roots based on business process management have remained the same. Nor has the vision that Matt Calkins, CEO of Appian, offered us a few years ago in this interview changed.

Even back then, he was talking about generative artificial intelligence before the term had even been coined in the market, which made the headline of that interview much easier for us: ‘In the near future, there will be almost no difference between talking to a computer and talking to a friend.’ He hit the nail on the head.

A little later, during his speech at the Appian World 2023 conference, he put forward the concept of private AI so that it could be adopted at the enterprise level. Calkins has therefore been a visionary of what is finally happening in the technology market, something that has propelled Appian to become a global benchmark in enabling technology to be used by companies to become more efficient and profitable.

On this occasion, we were able to chat with Ramón Riquelme, Senior Advisory Solutions Consultant at Appian, who told us how the latest developments presented by the company during the Appian World 2025 conference, held at the end of April in the US city of Denver and which Silicon had the opportunity to attend, are being rolled out in Spain.

Appian has placed artificial intelligence at the centre of its processes, which are the seed and raison d’être of a platform that goes beyond the origins of BPM to control everything related to RPA, process mining and unified data management… with AI always touching every component of the platform in a cross-cutting way.

Below is our interview with Riquelme:

—To begin with, I would like you to tell us how your Appian Around the World Madrid event went. What did you show, and what interest did the professionals who attended show?

This year, the main theme was undoubtedly Artificial Intelligence. Nonetheless, distinct from what we have seen in previous years. The agents we presented are changing the rules of the game.

AI is no longer a ‘toy’ where the user occasionally asks a question and gets a useful answer on certain occasions.

Now, we are allowing AI to give us information about how decisions are being made or what we have done from a business perspective. We can find out about each of the actions taken by an AI agent and how decisions were made: why something specific was done, what the input and output data were, what sequence of steps it took, and even where it got the information from, the specific source that justified our decision.

All of this translates into enterprise-grade quality, which is why it has sparked the interest of many companies to see how far we can really go with AI.

On the other hand, the situation we see today is that there are so many different offerings for tackling parallel issues that we don’t really know how to differentiate between them in order to use one tool or another. This problem puts us in a privileged position.

—You mean a platform capable of unifying and integrating all business processes and information so that AI can do all that work…

That’s right. Our platform not only has artificial intelligence, but also the potential to integrate with other things, the potential to provide tools to agents. And what tools do we provide to an agent? The ability to perform many tasks, which are precisely the ones we used to do on a regular basis.

From integration to starting a process that at a given moment executes an action or even coordinates something much more complex between a lot of different departments. That’s where we really have a lot to contribute.

On the other hand, staying within the same platform allows us to greatly reduce deployment, development, and definition times, etc. In short, it allows us to abstract ourselves from the complexity of integrating components.

—Over the years, the Appian platform has grown in terms of functionality, from BPM to RPA, process mining, data fabric, AI, etc. Today, which part of your platform, which is also modular, is most popular in terms of adoption?

In terms of adoption, there is equal interest in all the areas you mention. This is because artificial intelligence needs to be provided with tools, but also with information on which to base decisions and, finally, the structure and coordination so that everything as a whole can be presented to people.

I would emphasise the data part, because the information that organisations handle comes in most cases from dispersed systems. It is essential to be able to provide all the data that adds value. And here the first thing that is necessary is a great capacity for integration.

Appian makes it possible, with the Data Fabric plane, to interrelate all this information, so that it can be seen as a whole, not as scattered islands in different points.

Without that integration, you need to tell AI agents where the data they need to use is. By contrast, with Data Fabric, you simply tell the agent ‘this is your network of information already linked together to run your process’.

-Yet not everyone can have access to the same information, so security and granularity are crucial here. How does Appian deal with that?

Indeed, that is another point of great interest. It doesn’t seem like a good plan to leave all the information available for users to do whatever they wish.

So, on this network of information already related to each other, Appian explicitly segregates the security limits of each item. This way, when the user asks the artificial intelligence system a question, it will be able to do so with the specific context to which the user has access.

Data Fabric is able to segregate very well because it sets the necessary security limits at the moment of defining the data. In this way, when we stand in front of the screen, Pablo will see one thing and Ramón will see another, depending on the work and access granted to each one of us.

In other words, security is inherent in the data. It is not defined on the screen, nor in the agent or in the graph, but is defined during the integration of all the information.

-Moving to processes, Appian has excelled since its origins in the field of BPM. Yet the technology for designing and defining business processes has drastically changed and evolved during this time. Are the processes designed 10 years ago still valid, or has it been necessary to rewrite them to adapt them to new technologies?

Appian has always worked along the lines of trying to make its applications as ‘backwards compatible’ as possible. It is true that over time the way everything is displayed on the screen has changed, as is the case with the different versions of HTML. However, the definition remains the same and can continue to evolve to take advantage of new technologies as they emerge without having to touch practically anything.

Nevertheless, this is also where the current role of agents and process mining comes into play, because they are now able to track the original definitions and detect possible inefficiencies or improvements in an automated way. And our platform makes it very easy to replace parts with improved parts or parts that are able to do more with today’s technology. Moreover, if we find a performance improvement with those new parts, we can replicate them elsewhere or experiment with them dynamically.

-Finally, I would like to ask you about AI agents that communicate with each other without human need. How does Appian approach this new generation of artificial intelligence?

For me, and this is something you’ve also heard from our guru and CEO, Matt Calkins, the process layer is the key to making that coordination happen in a controlled way. The agents will have a set of tools, data, and defined objectives, but they will always have an input and an output.

At both ends, a process control layer must be put in place so that the agents can communicate with each other in a controlled way. They can be given some freedom to communicate, but always with well-defined barriers and rules in the process layer, which will allow them to respond to business objectives, regulations and the needs or requests of each organisation.