AI Is Changing The SDLC: Coding Matters Less, Specs Matter More

The traditional software development life cycle used to follow a fairly predictable breakdown of time spent on analysis, design, coding, and testing. Now that AI agents are capable of generating the code itself, those old percentages are being completely flipped upside down. In this episode, Ben Griswold (Grizen) and Noah Heldman (OutcomeSource) examine how AI is changing the daily reality of software delivery for both Greenfield (brand new) and Brownfield (legacy) projects. They discuss how a heavy reliance on upfront spec writing might actually be a return to Waterfall development, and debate the productivity differences between a small AI-assisted team versus a traditional enterprise pod. The conversation also covers the massive opportunity AI presents for modernizing decades-old COBOL and Fortran systems, along with a hilarious cautionary tale about a vengeful developer who used an early internet translator to sabotage a codebase. In This Episode, You'll Learn: How the traditional breakdown of software development time is shifting in the AI era. Why Greenfield projects now require heavily front-loaded analysis and design phases. The debate over whether spec-based AI development is essentially just Waterfall. How AI can safely untangle and modernize massive legacy systems using the Strangler Fig pattern. A funny story from the 1990s about a disgruntled consultant weaponizing German code comments. Why the demand for deep problem-solving skills will always outpace the demand for writing raw syntax. Connect with Us: Ben Griswold | Grizen: https://grizen.com Noah Heldman | OutcomeSource: https://outcomesource.com