AWS re:Invent 2025 - Move fast & don't break things: Maintaining software excellence as you adopt AI

Maintaining Software Excellence in the Age of AI

Defining Software Excellence

• The presenter, Ganesh, co-founder and CTO of Cortex, discusses the importance of maintaining software excellence as organizations adopt AI-powered coding assistants.
• He starts by defining what software excellence means, focusing on the top concerns of engineering leaders:
  1. Security and quality regressions
  2. The impact of these concerns on the business

Security and Quality Regressions

• Security risks include:
  • Secret leaks
  • Vulnerabilities and data breaches
  • Particularly concerning for consumer-facing businesses where customer trust is critical
• Quality regressions manifest as:
  • Increased incidents and SLA/SLO breaches
  • Higher costs of maintaining the business due to more time spent on incidents, bugs, and escalations
  • Ultimately leading to customer impact and loss of trust

The Impact of AI on Software Development

• AI is acting as an "amplifier" in the software development lifecycle:
  • AI is writing more code, but developers understand it less
  • Increased volume of pull requests leads to less thorough code reviews
  • Ownership and accountability for the code become unclear
  • Incident management becomes more challenging due to lack of understanding

Mitigating Risks with Proven Practices

• Focus on foundational engineering practices:
  1. Code quality and testing:
     • Write more tests to define guardrails for AI-generated code
     • Set up monitors and SLOs to detect regressions
  2. Security practices:
     • Secure all repositories, not just the "active" or "critical" ones
  3. Maintain human accountability and ownership:
     • Establish clear ownership and accountability for services and code
     • This drives the right behaviors, like investing in testing and quality

Adopting AI Coding Assistants with Confidence

• Build on a stable foundation of engineering best practices:
  • Invest in CI/CD, production readiness, and security processes
• Measure the impact of AI coding assistants on key metrics:
  • Focus on customer-centric metrics like incident rates, SLO breaches, and cycle time
  • Avoid getting too caught up in adoption metrics alone
• Make the most of AI coding assistants:
  • Leverage tools like internal developer portals to drive AI readiness, measure impact, and scale best practices

The Role of Internal Developer Portals

• Internal developer portals (like Cortex) can help organizations:
  1. Drive AI readiness through scorecards and best practice enforcement
  2. Measure the impact of AI coding assistants on key business metrics
  3. Facilitate the adoption of AI maturity practices across the organization

Key Takeaways

• Maintaining software excellence in the age of AI requires a focus on foundational engineering practices, not just adoption of new tools.
• Security, quality, and ownership are critical concerns that must be addressed as AI becomes more prevalent in the software development lifecycle.
• Internal developer portals can play a crucial role in helping organizations adopt AI coding assistants with confidence by driving readiness, measuring impact, and scaling best practices.