AIDevelopmentProductivity
AI-Assisted Development: What It Actually Means for Your Project
Brian Foley
Beyond the Hype
AI-assisted development isn't about pushing a button and getting a finished application. It's about amplifying skilled engineers so they can deliver higher-quality work in less time. Think of it as power tools vs. hand tools — you still need a skilled carpenter, but they can build faster.
Where AI Excels
In our modernization work, AI tooling delivers the most value in these areas:
- Code translation — Converting legacy code patterns to modern equivalents. AI is remarkably good at translating VB6 to C#, COBOL to Java, or PL/SQL to application-layer TypeScript.
- Test generation — Given a function and its expected behavior, AI can generate comprehensive test suites in seconds that would take hours to write manually.
- Documentation — AI can analyze undocumented legacy code and generate clear, accurate technical documentation.
- Code review — AI catches bugs, security vulnerabilities, and performance issues that human reviewers might miss.
What AI Can't Do
Architecture decisions, stakeholder communication, understanding business context, and making tradeoff judgments still require experienced human engineers. AI is a tool — and like any tool, it's only as good as the person wielding it.