Operating across 19 tribes and 130 squads, Localiza faces the fundamental challenge of scaling generative artificial intelligence without losing operational control. Rather than deploying isolated point solutions, Localiza organized its collaboration with OpenAI around three strategic pillars: transforming the business through experimentation at scale, creating new customer interfaces, and building AI-native products. For the company, generative AI operates as a core computing interface rather than a mere productivity add-on.
This approach directly targets the efficiency of internal operations by distributing ChatGPT licenses across every department, ranging from finance to day-to-day operations. Teams maintain active experiments with custom GPTs, agents, and Codex. While Codex initially landed within the technology department—engaging software engineers, product managers, designers, and data scientists—adoption rapidly expanded to analysts, managers, PMOs, and lawyers who began building their own prototypes and automations.
“We see this as a powerful tool that helps our teams do what they already do well. They live our culture, understand our customers’ needs, and know how to deliver great service.”
This broad distribution of tools changes the internal product pipeline. Multidisciplinary teams now prototype solutions before software development officially begins, shifting the primary operational bottleneck from technical execution to human capability development. The central challenge lies in helping a large workforce acquire new skills while keeping all 19 tribes and 130 squads operating efficiently and consistently.
Integration into the ChatGPT Ecosystem and New Acquisition Channels
As consumer behavior shifts toward natural language queries, language models have turned into a primary acquisition channel. Localiza’s internal data confirms that traffic originating from LLMs has grown rapidly since 2024, with ChatGPT emerging as the main channel for intent capture. To capitalize on this shift, the company integrated an application directly into ChatGPT in March 2026, becoming Brazil’s first mobility platform and one of the first in Latin America to execute such a release.
Through this integration, customers can explore rental, subscription, and used-car options directly within the conversation interface before completing the transaction in Localiza’s secure environment. The company treats this channel as an active experiment designed to track customer behavior at the frontier and accelerate operational learning from real interactions.
Internally, the transition toward AI-native products materializes in Liza, a conversational assistant built on a modern architecture that replaces rigid decision trees. Liza combines OpenAI models, Localiza services, and specialized tools around conversation, intent, and context. How does this architecture handle customer journeys? The system manages repetitive requests from initial inquiry through key stages of the rental process, freeing human teams for complex interactions.
Localiza monitors conversion rates and customer experience metrics, building on a foundation of service journey mapping established prior to the deployment of modern AI tools. Aggregated signals from interactions feed into a continuous learning loop supported by voice-to-text and text-to-voice capabilities.
Strategic Conclusion of the Transformation
The economic viability of large-scale LLM deployment depends entirely on disciplined operational integration and a clear focus on the customer experience. Localiza demonstrates that scaling generative intelligence requires systematic restructuring across dozens of organizational squads rather than isolated budgetary experiments.
Against the field
Moving the showcase for car rentals, subscriptions, and used vehicle purchases directly into the ChatGPT chat window demonstrates how Localiza is reshaping the initial customer touchpoint. Rather than relying solely on traditional software, the company embraces a third-party conversational interface while distributing AI licenses to legal and project teams for internal prototyping. Across all 19 tribes, legacy decision trees have given way to the Liza assistant, which handles repetitive customer inquiries throughout the rental flow. Will such extensive experimentation with OpenAI licenses across 130 teams translate into a measurable boost in conversion rates?