Intent-Driven Infrastructure at a Leading Latin America Bank

This leading Latin America bank has served its home market for more than a century and is today one of the region's largest financial services organizations, spanning banking, insurance, and digital services for millions of customers. While it maintains an extensive branch network, the most important customer transactions now run through its digital channels - and behind that experience sits one of the largest technology implementations in Latin America. The scale is serious: several thousand engineers work on the bank's platform, operating more than 1,000 applications and pushing thousands of infrastructure deployments every month. The bank runs on multiple clouds. Its engineering organization is executing a company-wide digital transformation with developer experience at its center: give developers simplified, self-service access to infrastructure so product teams can ship at maximum velocity —without compromising the security, governance, and compliance obligations of a regulated financial institution.

Developers now provision infrastructure two ways: through the Backstage self-service portal, or in natural language directly from AI IDE tools like Claude Code and GitHub Copilot via the StackGen MCP Server. Either path, StackGen translates intent into hardened, compliant OpenTofu — complete environments in minutes instead of days of waiting. Reusable, modular infrastructure components (L1/L2 modules) deploy consistently across environments, with centralized state management, module lifecycle management, and full Git (GitLab) traceability. StackGen's built-in policy engine validates every deployment against the bank's security and regulatory requirements in real time — guardrails first, not review queues — while AI-enabled context over the infrastructure landscape gives architects the visibility to make data-driven optimization decisions proactively. StackGen also converted the bank's existing cloud-specific templates to OpenTofu, completing the move to an open, cloud-agnostic IaC standard.

Application deployment that previously took six months through traditional processes is now delivered in weeks. For developers, the bank projects around 40% efficiency gains from reduced infrastructure connected toil; for the platform team, up to 10x less manual infrastructure work as TicketOps-style provisioning is automated away, with 25–35% productivity gains across developer and platform teams. Over five years, the bank estimates a 30% reduction in engineering effort on infrastructure — capacity it is deliberately redirecting toward AI initiatives and productivity — alongside a projected 270% return on investment, including a 50% reduction in security-incident and compliance-related costs. The rollout has two steps: first, scale these capabilities to hundreds of developers; second, implement agentic AI for DevOps, SRE, and infrastructure on a single reference architecture. The bank will use agents to connect policy, intention, execution, and verification so that autonomous operations stay safe and auditable in a banking ecosystem. The destination is intent-driven autonomous operations: a developer describes what they need and gets a secured, validated, verified result minutes later.