Details
- NAB evaluated Amazon Q and GitHub Copilot before standardizing 6,000 developers on Cursor, achieving 3x faster legacy modernizations with higher quality; expanding to over 10,000 employees including engineers, product managers, and designers.
- Key players: National Australia Bank (NAB), Cursor AI coding platform; engineers like Coby Paterson, Harjot, Chris De Lorenzo; NAB's Head of AI Tooling & Delivery Caroline Trang.
- Cursor excels in model flexibility for cost/latency tasks, superior codebase understanding across Java/React/COBOL/Assembly repos, and extensibility via NAB CEL library using rules/skills/hooks for standards and guardrails.
- Previously stalled projects accelerated: BizCalc monolith to Java/React microservices in 2 months vs. 6; Assembly mainframe migrations 3x faster; hardware-agnostic Kotlin payment app in 3 weeks vs. 4 months, with 5-8x velocity gains.
- NAB NAB News confirms 6000+ using Cursor/Q Developer with 40x requirements generation and 5-6x development uplifts; Cursor's 2026 study shows 68% surge in complex tasks across 500 teams, validating enterprise AI coding shifts.
Impact
NAB's 3-8x productivity gains in legacy migrations and greenfield apps signal accelerating enterprise adoption of agentic AI coding tools, outpacing rivals like Amazon Q and Copilot. This boosts developer focus on architecture (68% complex task surge per Cursor study), reshaping R&D toward AI-native workflows. Over 12-24 months, expect surging funding for Cursor (seeking $50B+ valuation) and competitors, driving on-device inference and multi-agent systems in banking tech stacks.
