On a Friday evening last April, just ahead of Easter, Argentina’s Central Bank stunned markets by announcing the end of nearly six years of strict currency controls, lifting restrictions on the purchase of US dollars. As Argentinians absorbed the news, staff at Galicia, one of the country’s largest private banks, braced for a hectic weekend.
Galicia’s technology team sprang into action to enable free dollar transactions in the bank’s app before business resumed on Monday. One factor proved decisive: They used AI to expedite coding for the new feature.
Developers at Galicia leveraged GitHub Copilot, an AI developer tool that empowers teams with AI across the software development lifecycle, to deliver the updated app on time.
“Code assistants help us gain the speed we need to make these kinds of things a reality… and to do it with the necessary security checks and the quality it deserves,” says Germán Gross, head of architecture at Galicia, who oversees the design and implementation of the bank’s technology systems.
Marcelo Fabián Gutiérrez, a senior developer at Galicia, calls the turnaround “highly unlikely” without GitHub Copilot, which he relies on almost daily. “No doubt about it,” he says, flashing a lopsided grin. “You save time, and that extra time leads to improved work performance.”
He adds that Copilot has made him more productive. “It’s giving teams and developers a lot more agility, which really helps us tackle more tasks.” The tool enables developers to generate full lines of code or complete functions through a variety of features, like context-aware suggestions, natural language prompts and real-time autocompletion.
Across Argentina’s financial sector, developers are increasingly using AI to write and test high-quality code faster, enhancing their own expertise and delivering better services to consumers. Many say they are learning from AI too, discovering creative solutions to resolve code problems that they hadn’t considered before.
Companies benefit as well, reaching their goals faster, attracting more customers, and accelerating technological growth – key advantages in the competitive financial industry, the tech leaders say. Bank app users enjoy quicker feature updates and improved reliability.
The financial sector – regionally and globally – is undergoing rapid transformation, with new digital banks entering the market and traditional banks adapting fast to avoid losing market share. In Argentina, a long history of inflation, currency controls and economic uncertainty have shaped an even more complex environment.
Galicia
Improving the experience
GitHub Copilot is built on GitHub, the leading developer platform to build, scale and deliver secure software, and integrates with popular code editors like Visual Studio, Xcode and Neovim. By the end of 2024, GitHub reported nearly 1.2 million developer accounts in Argentina.
GitHub Copilot offers three real-time chat modes: one for questions and code suggestions, another for editing and a third agent mode that can manage multistep tasks like building features or creating files across an entire project. Additionally, developers can delegate routine and specialized tasks to the GitHub coding agent to move projects forward, even while they’re offline.
Powered by the latest generative AI models, GitHub Copilot prioritizes security, privacy, compliance and transparency. The tool encrypts data in transit and at rest on Microsoft Azure.
Galicia, which is part of Grupo Galicia, has over 5 million clients and adopted AI code assistants in early 2024 as part of a broader digital transformation. Today, around 500 developers at the bank actively use GitHub Copilot. “You simply can’t succeed as a bank today without embracing an innovative mindset,” Gross says. “It is mandatory.”
Founded in 1905, the bank’s core business has long centered on offering loans and managing deposits to medium- and high-income people and businesses.
At Naranja X, a fintech also under Grupo Galicia, the adoption of GitHub Copilot was in its DNA. Specializing in digital credit services, Naranja X manages more than 13 million credit and debit cards.
Sofía Baggini, DevOps and automation manager at Naranja X, says developers save two to three hours daily with GitHub Copilot. Time spent troubleshooting specific tasks dropped by 50%, she adds. A back-office developer even used Copilot to fix front-end connection problems outside of his usual expertise.
“The quicker the product reaches customers, the happier they’ll be and the better their interaction,” Baggini says. “If a click takes one second instead of five, that impacts the experience.”
Javier Blanco, a Naranja X developer focused on automated testing, uses GitHub Copilot to suggest code, fix errors, and test final products. He calls AI “a very good accelerator” that helps him juggle multiple projects simultaneously, get higher quality, find creative solutions and switch seamlessly between programming languages.
“It’s like becoming bilingual, but in programming – you’re learning a whole new language,” Blanco remarks.
More than 80% of Naranja X developers – approximately 560 users – are now actively using GitHub Copilot, according to Pablo Fernández, the company’s tech lead. “The impact is significant since access was granted to all; anyone can request and automatically receive a license without paperwork.”
Naranja X
Less time, more value
Gutiérrez, the Galicia developer, used to spend hours searching blogs and manuals for coding solutions. Now, he simply asks Copilot. For a recent anti-fraud feature for the Galicia app, GitHub Copilot improved his draft code and flagged potential issues. He then used the assistant to create unit tests to confirm the feature worked.
By leveraging GitHub Copilot, developing comprehensive test cases for this project took 60% less time. “For me, it adds real value – this is nothing short of a revolution,” Gutiérrez points out.
An internal survey at Galicia found that about 80% of developers feel more productive with Copilot, says Gross. His team of developers utilizes the tool across its various modes, including the coding agent.
The head of architecture emphasizes that developers remain essential as AI can be inaccurate at times or miss context. In fact, he believes AI may increase the demand for new software features. AI code assistants “are like steroids for any software developer,” Gross observes.
Gonzalo Manuel Calderón, another Galicia developer, was initially skeptical. Now, he uses GitHub Copilot to autofill code directly in the editor. After compiling the text, he uses the chat to spot errors and create unit tests. “It’s like getting a third opinion,” he says.
With Copilot’s help, he built a mortgage redirect page from the personal banking section in just two days—compared to the usual one-week timeline. He’s also finishing a computer engineering degree at the University of Buenos Aires and uses Copilot as a learning tool for his studies.
“It’s a lot like interacting with a teacher,” Calderón adds, pointing out that, just like with people, the more context you give AI, the better its responses. “It’s going to transform how we work.”