BNP Paribas Cardif plans in-house AI agents for Brazil

BNP Paribas Cardif will develop AI agents in-house for Brazil, where half of claims decisions are automated and 40% of claims processes need no human intervention.

Wire notes

  • BNP Paribas Cardif plans to develop AI agents inside its French group for Brazil’s next growth cycle.
  • About 40% of claims processes require no human intervention, while half of claims decisions are automated.
  • Operations and technology headcount has remained nearly unchanged for three or four years, Silva said.
  • Only one in five AI models could sustain itself in a survey Silva cited.
  • Fernanda Campos said 78% of employees completed more than seven hours of AI training.
Sem recorrer às big techs, seguradora francesa prepara agentes de IA para crescer no Brasil
PhotoSem recorrer às big techs, seguradora francesa prepara agentes de IA para crescer no Brasil — geralt / pixabay

By Amara Osei

BNP Paribas Cardif says about 40% of its insurance claims processes now need no human intervention, while the insurer automates half of claims decisions. Cardif cautions that not every automated process uses artificial intelligence.

The French insurer plans to incorporate internally developed AI agents into its next growth cycle in Brazil. It will build the models within the French group, not use third-party big-tech solutions, to speed implementation. Technology and operations director Alex Silva and Emmanuel Pelège, CEO of the Brazilian operation, said the strategy advances AI sustainably, with financial returns and data security as priorities.

What will the in-house agents change?

Cardif has not disclosed which tasks the agents will handle, when they will launch, what they will cost or how it will measure their return. The commercial case therefore remains undefined at this planning stage.

Pelège said: “I think this efficiency is used to finance our transformation. And also to really improve the partner and customer experience.” He spoke after being asked about AI’s effects on operations and possible job cuts.

His statement links efficiency to transformation funding and service improvement, but it does not quantify either benefit.

Takeaway: Cardif wants more autonomous software, but its first use case and launch date remain undefined.

How much claims work already runs without people?

Those figures do not make the operation an AI-only system. Cardif already uses machine learning (algorithms that identify patterns in data) in claims analysis, but the company says not every automated process relies on artificial intelligence.

Silva said the number of employees in operations and technology has remained practically unchanged for three or four years despite business expansion. He attributes that stability to investment in automation.

Silva said: “We are using technology to sustain our growth at scale. [...] In the next cycle, we want to put the agent models to work for our benefit and sustain growth without increasing costs.”

That is a company claim. The information provided contains no absolute staffing figures or cost series with which to test the link between automation and stable headcount.

Takeaway: Flat staffing is consistent with the automation story, not proof that AI caused the result.

What must the agents prove before scaling?

Silva identified two barriers to adopting the technology at scale:

  • Turning experimental projects into applications that generate financial returns.
  • Avoiding exposure of sensitive information.

He cited a survey presented at an international event in which only one in five AI models could sustain itself. The survey cited difficulty gaining scale and a lack of return on investment as reasons models fail.

The event, sponsor and methodology were not identified, so the one-in-five figure cannot be independently assessed from the information provided.

Silva also said many companies come under attack when they try to accelerate models without adequate sustainability in information security. The statement did not specify the nature of those attacks or identify examples.

Cardif has not disclosed the business-case target, deployment budget or security controls for the planned agents.

Takeaway: The business case requires measurable returns without exposing sensitive information.

How is Cardif preparing employees?

Fernanda Campos, the insurer’s human resources (HR), environmental, social and governance (ESG), and communications director, said 78% of employees have completed more than seven hours of AI training through an academy Cardif created.

That figure measures training exposure, not proficiency, role redesign or job security. Cardif cited the program amid questions about possible layoffs. It did not announce job cuts or identify positions that agents would replace.

Takeaway: Cardif’s next test is whether its in-house agents can deliver measurable returns, protect sensitive information and sustain Brazilian growth without adding cost.

via oglobo.com (Source)

Filed under

Share this article:

More from Amara Osei

Amara Osei

Show full bio

News editor covering people, launches and day-to-day practice in marketing, advertising, marketing technology, media, SEO and search at Marketing Herald.

354 articles

More on the wire

« Previous article