From two hours to 15 minutes

For DANA Indonesia, building a cashless society means delivering a payments experience that works even on low-bandwidth connections. That leaves little room for buggy releases that create real user repercussions. In 2025, DANA’s engineering teams used O’Reilly to build AI agents that catch errors before they ship, with results that speak for themselves.

Learn about O’Reilly for Teams Explore other success stories

Working together

Challenges and solutions

The challenge

  • 53% Android patch rate 53% of Android releases and 39% of iOS releases in 2024 required patches, a significant problem in a low-bandwidth market
  • 156,000-line test gap Legacy code with insufficient unit test coverage across more than 156,000 lines
  • Hours lost per task Slow feature development cycles with ramp-up times of one to two hours per task

The O’Reilly solution

  • AI-powered pull request review Engineers used O’Reilly to build prompt engineering foundations, then deployed AI agents to automatically review pull requests for known patch-causing errors
  • Legacy code test generation Web team used O’Reilly to apply solutions for unit test generation across legacy code
  • AI-generated feature code Frontend team piloted AI programs to auto-generate feature code

Results

Here’s how investing in learning translated into measurable business results.

  • 53% to 33% of Android releases required patches, with critical issues caught by AI agents before production
  • 79% to 82% web module unit test coverage in just two months
  • 83% reduction in code development time, from one to two hours to 15–20 minutes

Takeaway

DANA Indonesia put AI agents to work on real code, in production, with results users can feel

O’Reilly gave them the foundation to ship better code, faster, and a structured roadmap for what comes next.

Ready to put AI to work in your engineering teams?

Talk to us