By InnovAction Editorial – War Room
Interview with Francesca Saraceni – CEO and Co-founder of Intellico, Adjunct Professor @GSoM
In the past two years, generative artificial intelligence – such as the technology powering tools like ChatGPT – has captured widespread media attention. Yet beyond the buzz and bold predictions, more pressing questions arise: what is its actual impact on the economy? And more importantly, how is it truly being implemented within companies?
In the October 29, 2024 interview with Luca De Biase on the program War Room InnovAction, Francesca shared her perspective. This article highlights the key insights that emerged from her contribution.
The Four Waves of Artificial Intelligence: A Guide to Navigating the Future
For Francesca, the first step in understanding AI begins with having a clear map. The framework she uses is inspired by Kai-Fu Lee’s 1 model of four waves, which helps distinguish areas, technologies, and levels of maturity:
- Internet AI
This is the first wave, based on the massive use of available online data. It includes recommendation systems, search engines, e-commerce platforms, and more recently, generative language models. In this phase, AI analyzes preferences, behaviors, and content to offer personalized suggestions. It’s the AI embedded in our daily lives, always just a tap away on our smartphones. - Business AI
The second wave brings AI inside organizations, working with internal data: orders, sales, technical drawings, management reports. The goal is not just to automate, but to support and enhance decision-making. This is the area where Francesca, with Intellico, works most closely. Here, AI becomes a colleague at the service of all company functions, from design to logistics, from research and development to marketing. - Perception AI
This is the intelligence that “hears” and “sees”. It bridges the gap between the digital and physical worlds using cameras, microphones, sensors, and voice interfaces. When you order food with a voice command from your living room or unlock your phone with Face ID—are you engaging with the physical world or the virtual one? In this wave, AI interprets the environment to create seamless human-machine interactions. - Autonomous AI
This is the most advanced frontier – and also the most debated. It refers to intelligence that has not yet been conferred upon machines but is the subject of research by scientists and scholars worldwide. It is the intelligence that most concerns us, as it is closest to the idea of artificial intelligence that can fully replicate human intelligence. The goal in this case is to provide systems with true decision-making autonomy. This is the AI of autonomous vehicles, robots moving in space, and systems capable of adapting to changing environments. A challenge that requires the integration of all the previous waves.
Autonomy here does not only mean automation but the ability to reason, make decisions, act in the real world, interpret principles, and project towards consciousness.
Business AI: Artificial Intelligence as a Tool for Business Transformation
Business AI is where the most tangible impact is being made for companies and organizations, whether public or private. Here, artificial intelligence is no longer theoretical but a working tool. And the results are tangible. Let’s revisit some of the most interesting examples shared during the interview:
- Industrial Design
In the metalworking sector, for example, companies that produce on demand must respond quickly to new customer requests. In this context, AI can analyze existing technical drawings to identify similar past designs. This enables up to 80% reuse of previous work in some cases, significantly cutting down design time and accelerating delivery. - Product Formulation
In the chemical, cosmetic, or food industries, the approach is similar: AI scans the archive of tested recipes to find formulations similar to those requested, avoiding duplication and accelerating research and development. - Production Planning
In high-uncertainty environments, AI expresses its full potential. Traditional tools, based on deterministic rules, cannot manage complexity. AI, however, uses probabilistic models and neural networks to build adaptive scenarios and optimize resources and processes even in dynamic conditions.
“These tools are not merely aimed at improving efficiency,” Francesca points out, “they are solutions that directly contribute to improving the business model and supporting strategic decisions, allowing tangible improvements on the company’s bottom line.”
In these contexts, it becomes even more important to provide and require “Explainable AI” solutions. When it comes to implementing AI solutions, resistance is typically significant: there’s a fear of human replacement and the concern of not understanding the reasoning the system used to make a certain prediction. Understanding how and why a system arrives at a recommendation is critical – even more so in B2B environments, where each decision can have wide-reaching consequences and involve entire teams or company functions.
“When we founded Intellico,” recalls Francesca, “we knew that trust in AI tools, their comprehensibility, and transparency would be key prerequisites for adoption at the corporate level.”
For example, a production manager or marketing head cannot fully take responsibility for a production plan or marketing campaign suggested by a non-transparent digital tool.
Explainable AI allows the recommendation to be accompanied by an explanation of the factors – the drivers – the algorithm considered to reach that result.
Having these explanations means not only receiving a prediction but also understanding why and how it was made. In other words, it provides more transparent, reliable, and usable decision-making tools.
The Quality of Data: The Key Factor for AI Success
All of this, however, depends on one critical foundation: data quality. While algorithms play a key role, it’s the data that carries the real value.
Looking ahead to the next five years, many of today’s AI tools—now seen as cutting-edge and unique—are likely to become standard features in the corporate tech stack. In this evolving landscape, where will the real competitive edge lie? The answer is clear: in the data.
Companies that know how to build, organize, and leverage their information assets – both explicit and implicit knowledge – will be the best equipped to tackle future challenges. The more this information is intelligently accessible, through advanced AI tools, the more the organization will be able to adapt, innovate, and make better decisions.
The real difference will lie in the ability to transform data into useful insights, integrating technology and human intelligence in an increasingly fluid and dynamic ecosystem.
“Many Italian SMEs,” explains Francesca, “are only just starting to structure the collection and management of their data. This is a mandatory step. Even though we’re starting from a place of significant technical debt, there seems to be a shift toward investment and targeted actions to increase the maturity level of our companies. Large companies started earlier, but there is still a need in this segment to approach data management with a more strategic and less tactical vision. In this sense, AI helps prioritize the areas that need to evolve toward organized data management.”
“Make or Buy”? How to Choose Between In-house AI Solutions and Ready-to-Use Solutions
Many companies ask themselves: is it better to develop an AI model in-house or purchase an existing solution? We suggest avoiding binary choices; the answer lies across multiple levels:
- Infrastructure
This is the level where companies are most likely to opt for “buy” solutions. Data centers, cloud computing, IaaS (Infrastructure as a Service) environments are now established, and often the costs and management complexities of replicating them internally do not justify the benefits.
- Models and Libraries
This is the first level where there’s room for a hybrid approach. One can start with open-source models and adapt them to their needs or use pre-existing models and libraries without necessarily reinventing the wheel. The choice depends heavily on the application area and the business goal: in some cases, developing makes sense; in others, integration is more effective. - Data
The real added value is in the company’s data. Depending on the needs, companies can decide whether to integrate with external data, involve partners, or keep everything in-house. Once again, the specific application area determines the best approach.
“Every company needs to ask the right questions,” emphasizes Francesca: “What is my competitive advantage? Do I have unique data? Do I have the expertise to customize a model? Am I ready to invest in AI governance?”
Conclusions: Artificial Intelligence as a Lever to Amplify Human Intelligence
At its core, the message is simple: artificial intelligence isn’t here to replace us—it’s here to enhance us.
For this reason, it is preferable to refer to AI not as Artificial Intelligence, but as Augmented Intelligence. The focus is on enhancing human decision-making and creativity by introducing additional layers of intelligence that complement—rather than compete with—human capabilities.
For those who know how to use it, AI is already a tangible reality. But unlocking its full potential requires three essential ingredients: high-quality data, accessible and intuitive tools, and a corporate culture that embraces change.
The real challenge isn’t technological—it’s organizational. And Italian companies, with their heritage of innovation and creativity, are well-positioned to meet it—provided they’re willing to invest in their own transformation.
References:
- Lee, K.-F. (2018). AI Superpowers: China, Silicon Valley, and the New World Order. Houghton Mifflin Harcourt.
