Loading
Description
According to Rômulo Cioffi, Chief AI and Innovation Officer at Squadra, the market needs to move beyond the idea that AI will be able to solve all problems and advance towards a more rational and strategic adoption.
Transcript
After an initial phase marked by pressure to experiment with artificial intelligence in different parts of organizations, companies are beginning to face a more pragmatic question: how much does the technology cost and what results does it actually deliver to the business? For Rômulo Cioffi, Chief AI Officer (CAIO) and COO at Squadra, the market needs to move beyond the idea that each new AI evolution will be able to solve all problems and advance towards a more rational and strategic adoption.
The executive spoke with TI Inside during an event promoted by Squadra and Sensedia to discuss a study conducted with the support of MIT. The meeting brought together about 40 leaders from different sectors in São Paulo and had as one of its main discussions precisely the transition from small experiments to applications capable of producing concrete results for companies.
In Cioffi's assessment, the popularization of artificial intelligence created intense pressure for adoption. The technology initially advanced in people's daily lives and later gained space within organizations, also accompanied by strong market exposure.
The result was a race to experiment. The problem arises when companies start calculating the return on these initiatives and discover that certain applications did not deliver what was expected.
Cioffi assesses that some organizations still do not clearly know how to start using artificial intelligence, how to measure its use, how much it costs, and whether a given use case is truly positive for the business.
The market is now entering a stage where these questions gain more importance. Instead of AI applied indiscriminately, the discussion now involves strategy, measurement, and impact.
The advancement of agentic AI has added a new stage to this dynamic.
Cioffi compares the moment to a gold rush. After the popularization of artificial intelligence models and interfaces, agents have taken center stage as a new technological promise.
For the executive, however, the agent needs to be understood as just another technology component capable of performing activities on behalf of the user, and not as a universal answer.
The problem arises when each new concept is pursued by organizations as if it were capable of solving all existing challenges on its own.
In the assessment of Squadra's CAIO, there is still excessive optimism in part of the market, and companies continue to invest resources inadequately because they do not start from a more rational analysis of where the technology can truly generate results.
The transformation does not happen only in the technological architecture. For Cioffi, one of the biggest impacts needed to democratize the use of AI within organizations lies in revising the role of people.
Professionals are increasingly taking on a knowledge curation role, evaluating whether artificial intelligence has correctly understood what was requested and whether the produced result effectively meets the defined objective.
This change requires a different stance from those who were previously focused directly on executing the activity.
At Squadra, the adaptation process began approximately three years ago and continues daily. The reason is the very speed of technological evolution, which demands constant learning.
The company works with groups at different levels of depth. Professionals with more technical knowledge study the technology in greater depth and develop more complex applications. The knowledge acquired by this group is subsequently shared with professionals who are at another stage of use.
People can also migrate between these groups as they develop new capabilities.
The intention is to create an environment where professionals learn and teach continuously, so that evolution is no longer concentrated in a few specialists and begins to feed the organization's culture.
In the vision presented by Cioffi, democratization depends on governance, and the organization needs to know which tool is being used, who is using it, who trained that person to use it, how much that activity is costing, and what result was produced.
Only then is it possible to evaluate whether a given knowledge or application deserves to be expanded to other parts of the company.
The accumulated experience itself needs to be transformed into a knowledge base. Paths tested and deemed inadequate do not need to be repeated by different teams, while practices that have demonstrated the ability to extract value at a lower cost can be shared.
Governance thus also functions as a mechanism for organizing the learning produced during experimentation.
The ultimate goal remains related to productivity and cost reduction, but the company seeks to build a culture where these goals are pursued with visibility into what is being done.
This process also requires space for initiatives that do not work.
Cioffi advocates for a culture where experimentation is part of everyday life and stopping a process that is not yielding the expected results is not necessarily treated as something negative.
Mistakes are part of discovery. When a team identifies an approach that works, the learning can be shared and celebrated as a way to achieve the result more efficiently.
For the executive, discussions among leaders from different organizations also help in this process because they show that adoption, training, and return difficulties are not necessarily restricted to a single company.
Cioffi also identifies a difference between artificial intelligence and previous technological waves.
Technologies like mobile phones, the internet, and cloud computing went through adoption cycles with greater differences between markets. In the case of AI models, availability happens much more simultaneously globally.
This means that organizations from different countries are trying to discover, at the same time, how to use the technology within their businesses.
For Cioffi, the study discussed during the meeting helped precisely to separate applications that are still surrounded by hype from those that are beginning to show greater consolidation.
The conclusion presented by the executive is that the process is still in its early stages.
New models and applications will continue to emerge, requiring companies to adopt a permanent learning stance rather than perceiving that the transformation has already reached a definitive stage.
The need for human participation also remains even with the increase in model capabilities. Cioffi believes that artificial intelligence and people have different characteristics and need to be applied according to what they do best.
AI can work on large volumes of data, recognize patterns, and produce responses based on the training received. Humans, in turn, remain responsible for capabilities related to creation, innovation, emotion, prioritization, and evaluation of the impact produced.
Therefore, the executive does not see people's roles disappearing as technology advances.
The professional becomes responsible for guiding the process, curating, evaluating whether a given result is adequate, and deciding where there are opportunities for improvement.
In the view of Squadra's CAIO, finding the right combination of these capabilities will be an important part of innovation. The challenge is not just to develop new models, but to discover where and how to use artificial intelligence within journeys so that it effectively produces transformation.
This search also requires inverting a common logic in the early years of generative AI. Instead of starting with the technology and later looking for where to apply it, the organization needs to identify which problem it wants to solve.
Cioffi places the human at the center of this definition. It is the person who identifies the pain point, sets priorities, and evaluates what impact they want to produce.
Artificial intelligence then enters as a new lever available to solve that problem.
For the executive, treating it as a technology capable of indiscriminately solving any challenge creates precisely the kind of expectation that led some companies to invest without being able to prove the results later.
The evolution towards more mature adoption therefore involves reducing the mystery around AI and rationally analyzing where it makes sense.
Squadra's international experience also reinforced Cioffi's perception that companies are facing similar challenges.
The executive reported recent experiences in France and the United States and the company's participation in a hackathon held in Seattle the previous year. The team participated in a competition with 14 companies, with Squadra being the only Brazilian one, and finished in second place.
For Cioffi, although there is a more restricted group of companies developing the large models, the application of artificial intelligence is globally distributed.
Each organization knows its own challenges and can find different ways to implement the technology.
This creates space for Brazilian companies to also develop relevant practices, as long as they can combine the creativity that the executive attributes to the national market with greater planning and governance capacity.
The moment described by Squadra is a change in the quality of questions asked by companies. After experimenting with models, copilots, and now agents, organizations are beginning to need to demonstrate where the return is, how much each initiative costs, and which applications truly deserve to scale.
For Cioffi, the next stage requires balancing experimentation and discipline: allowing people to test and learn, accepting initiatives that don't work, quickly sharing discoveries, and building enough governance to turn this learning into results. AI ceases to be an end in itself and must prove which problem it can solve and what value it effectively delivers.
Available at: TI INSIDE