How one healthcare software provider is using the latest AI tools to improve care for millions of Brazilians — without losing sight of the human experience
This article was contributed by Antonio Espirito Santo, CEO of IDS, and Thiago Rodrigues, R&D Manager of IDS, with support from Hanna Ha, Corporate Knowledge Programs Lead at Volaris Group.
When IDS began its AI adoption journey, the question that guided the team was never simply about what the technology could do. It was about how technology could enhance the experience for the healthcare professionals and patients they serve.
Seven years and nine features later, that discipline has produced a suite of AI tools that are used across Brazil, and their leaders have gained valuable lessons about what responsible innovation in healthcare looks like.
It started with conviction
In 2019, IDS’s former owner returned from a trip to Silicon Valley with a clear belief: AI would define the future of software. For a Brazilian healthcare company operating in a complex, high-stakes environment, that conviction could have led anywhere. It led IDS somewhere deliberate.
Rather than chasing the technology, the team asked a harder question first: where can AI make a real difference in the daily lives of healthcare professionals and the patients who depend on them? Serving hundreds of municipalities and millions of citizens across Brazil, IDS viewed AI as an opportunity to improve decision-making at scale.
“At that point, no other company in our market was implementing AI within its products,” says Antonio. “We believed the timing was right — but only if we could deliver something that truly helped our customers, not just something that was technically impressive.”
For Thiago, who would go on to lead the R&D team building these solutions, that foundational question shaped everything that followed. “From the beginning, the conversation was never about what AI could do in theory. It was always about solving a real problem: where a clinician was losing time, where information was getting lost, where a small improvement could change the quality of a decision.”
Keeping humans in the loop has always been a deliberate design principle for us. Human judgment remains indispensable, particularly in healthcare.
-Antonio Espirito Santo, CEO, IDS
The first feature: A problem worth solving
IDS’s first AI feature was a patient chart summary for the public healthcare network — chosen because anyone who has worked in a busy public clinic would immediately understand the problem it solved.
In Brazil’s public healthcare system, a patient may carry more than ten years of medical records. Reviewing that history during a consultation is not just time-consuming — it is often impossible. Physicians are forced to make decisions without the full picture.
“We built an AI-powered summarization tool that analyzes a patient’s records, identifies the most clinically relevant information, and presents it in a single, easy-to-read view at the moment of care,” Thiago explains. “But the hard part wasn’t the technology. It was understanding exactly what a physician needed to see in thirty seconds, and what could safely wait.”
That first feature is now licensed to over 20 customers and has established the design principle that has guided every AI initiative at IDS since: technology should enhance human judgment, not replace it.
Nine features, one philosophy
Today, IDS has nine AI features that are either live in production or in active testing across its healthcare and social assistance solutions, including medical and social summarization, clinical decision support, automated triage risk classification, chronic disease risk prediction, and keyword extraction. Internally, IDS has also developed a customer care assistant, an automated database migration engine, and a multi-agent tender analyzer that helps the company evaluate public bids more efficiently and operate as a more effective business.
Their range of features has breadth, but not at the expense of discipline.
We do not add AI simply because it is technically feasible. We apply it where it can create meaningful and measurable benefits for the user.
-Antonio Espirito Santo, CEO, IDS
Thiago describes the process that makes that discipline operational. “Before anything moves into development, we conduct a structured assessment with our product and business analysts. We ask questions such as: ‘What is the daily impact? Does it reduce administrative burden? Does it fit the actual workflow the professional is already using?’ If we cannot answer those questions clearly, the feature does not move forward.”
Every feature that has shipped keeps the human in the loop. In their Chronic Disease Risk Prediction tool — which flags diabetes, cardiovascular, and cerebral risk between triage and consultation — the output is explicitly framed as an alert, not a diagnosis. In the Automated Risk Classification tool, the triage suggestion is a recommendation only; final decisions remain with nursing staff and physicians.
“Keeping humans in the loop has always been a deliberate design principle for us,” Antonio says. “Human judgment remains indispensable — particularly in healthcare.”
Thiago adds: “And that principle actually makes the technology better. When you design for augmentation rather than automation, you are forced to think very carefully about what information the human needs, when they need it, and how it should be presented. That precision improves everything.”
Building trust in a high-stakes environment
Introducing AI into clinical workflows requires more than good technology. It requires trust — and IDS has learned that trust is earned through results, transparency, and proximity to the customer.
“Trust in healthcare is earned through results,” Antonio says. “We make it clear from the outset that healthcare professionals are the experts. Our role is not to replace their judgment, but to help them work more efficiently and focus on what matters most.”
Thiago points to the development process itself as a trust-building mechanism. “Medical Summarization resonated quickly because it was built alongside the people who would use it. We spent significant time with clinicians during design and validation — not just testing the technology, but understanding what they needed to feel confident using it. That investment paid off.”
For customers who are more cautious, the approach is consistent: demonstrate, don’t argue. “Many healthcare professionals take a seeing-is-believing approach, which is completely understandable,” Thiago says. “Once they experience the benefits firsthand — saving time, accessing better information, improving the quality of care — many of their concerns naturally begin to fade.”
The honest lessons
IDS has not only shipped successes. IRIA, their customer-facing support chatbot, did not perform as originally envisioned. The complexity of IDS’s platform made it difficult to achieve the accuracy required for autonomous customer-facing interactions.
Rather than persisting with a model that wasn’t delivering, the team changed course. IRIA was repositioned as an internal tool for the Customer Care team — helping support analysts find information faster and resolve tickets more consistently, with a human expert still in control.
“That was an important lesson for us as a development team,” Thiago reflects. “We had built something technically sound, but the environment was more complex than we had modeled. The right response was not to push harder — it was to reassess where the technology actually created value and redesign around that.”
Antonio summarizes the broader lesson: “AI does not always create the most value through full automation. Sometimes the best outcome comes from augmenting human expertise rather than attempting to replace it. Learning when to shift from an autonomous model to a human-assisted model has been one of the most valuable lessons in our AI journey.”
Both leaders are candid about the other major challenge, which was cultural adoption. It proved harder than technical development.
“In the early stages, AI initiatives were concentrated within a small specialized team,” Antonio explains. “While that helped us move quickly and build expertise, it also meant the broader organization was not involved in the transformation of the business to the same extent. Looking back, we would have achieved faster results if we had engaged a wider group of employees much earlier.”
Thiago agrees, and sees it as an ongoing responsibility. “The technology can be implemented relatively quickly. Building confidence and familiarity across the organization takes longer. Today, we place much greater emphasis on education, communication, and participation when introducing new AI capabilities. It is as much a people initiative as a technology one.”
What comes next
When asked which initiative will matter most in three years, both Antonio and Thiago resist the temptation to name a single feature.
“At the moment, it feels like everything we are building matters,” Antonio says. “The real challenge is not identifying opportunities — it is prioritizing them.”
Thiago points to the underlying shift that makes that prioritization possible. “We have moved from AI being a specialized effort to AI being a company-wide capability. Different departments are now actively identifying opportunities in their own areas. That cultural change is as important as any single feature we ship.”
If there is one area both leaders highlight, it is decision support — the ability to help healthcare professionals access the right information at the right moment to make better-informed decisions for their patients.
“Three years from now, the initiative that matters most may not be a single feature,” Antonio reflects. “It may be the fact that AI has become a natural part of how healthcare professionals interact with information, make decisions, and serve patients across our entire platform.”
For a company that started with one question — how can we help our customers in ways that make a real difference? — that outcome would be a fitting answer.
