CursorCursorProblem & solution
National Australia Bank accelerates legacy migrations with Cursor
Summary
NAB standardized 6,000 developers on Cursor after evaluating Amazon Q and GitHub Copilot, and is expanding access to more than 10,000 technology employees. The bank uses Cursor across legacy modernization and new development: BizCalc's Silverlight/.NET monolith is being refactored into Java microservices and React, while Assembly-based core-banking programs are being migrated using generated flowcharts and business summaries. Engineers use Ask Mode, Plan Mode, custom NAB CEL skills, Composer, and Opus coding models to document systems, produce requirements and API specifications, parallelize implementation, and test changes. NAB reports that both modernization efforts are progressing three times faster than expected, while a hardware-agnostic payment app was built in under three weeks rather than four months; one team reported a 5–8x velocity improvement.
Context
NAB needed to modernize legacy systems, including a Silverlight/.NET monolith and Assembly-based mainframe programs, while resource constraints and limited expertise had stalled some work. Its merchant services team also wanted a hardware-agnostic payment application to avoid vendor lock-in.
Approach / What changed
NAB standardized Cursor across an initial cohort of 6,000 developers and built the NAB CEL context engineering library with Cursor rules, skills, and hooks. Teams used Ask Mode and Plan Mode to reverse-engineer systems and create requirements and API specifications, then used Cursor for implementation and testing, including Composer, Opus coding models, and parallelized subagent plans.
Takeaways
NAB CEL uses Cursor primitives including rules, skills, and hooks to centralize shared knowledge, enforce development standards, and set guardrails for agent behavior.
For the BizCalc migration, Cursor reduced pre-development work from an originally planned two months to one week; the full Java and React migration is expected to take two months instead of six.
The merchant services team built a hardware-agnostic payment app in less than three weeks using Composer and Opus after creating a plan for parallelized subagent work; the team reported a 5–8x velocity improvement.
Source and freshness
- Source
- cursor.com
- Published
- Added to Yomu
- Last updated
Summaries are generated from the original by a language model. Check the original before relying on it.
Keep reading
Related reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
The best engineering reads, once a week
One useful email. Unsubscribe anytime.