Expert Insights

Chris Rickard introduces an intriguing shift in software development with the infusion of AI, the return of the Waterfall model. Arguing against a one-size-fits-all approach, he demonstrates how AI changes the proportion of time spent on planning and executing, leading to better code quality.

Chris highlights the surprising revival of Waterfall, not in opposition to Agile but as an essential strategy when working with AI.

Listen to Chris explain:

  • How AI has shifted the bottleneck from coding to planning, reducing implementation time but increasing focus on specification.
  • Why detailed and precise planning leads to better AI outputs.
  • The significant role played by AI in dissecting legacy systems into detailed requirements.
  • How AI brings the possibility of turning significant codebases into functional specifications within hours.
  • The importance of trust in documentation, and how AI can help to keep it current and detailed.

Quote

quotation-marks icon
If it's a confident junior developer who wants to impress you and do everything you can, if you leave any ambiguity or any context out, it's gonna make an assumption and the assumptions are gonna fuck you up down on this other end. What does this new world look like when we can spend five minutes putting together 68 highly detailed requirements, and then we can spend an hour building that out? I think it's a new paradigm. The reason there was such a backlash against Waterfall was because companies would spend a year putting together a 500 page document. Then inevitably they'd find out after two years of development that it didn't work.quotation-marks icon
Chris Rickard ,
Founder, UserDocs

THE NEW DEFAULT angle

Here are relevant takeaways to guide software development teams transitioning to AI-assisted workflows:

  • Recognize where the bottleneck moves. As AI reduces coding time, more focus needs to shift to specification and planning. Adjust team roles and schedules accordingly.
  • Prioritize precision in planning. The more detailed and accurate the input, the higher the quality of AI outputs. Avoid ambiguity and provide clear context to ensure the best results.
  • Reconsider Waterfall as a significant planning approach when AI is involved. While Agile gives speed, a detailed, front-loaded planning stage can harness AI capabilities to the fullest.
  • Utilize AI for the requirement generation of complex systems. Deriving detailed conditions from large codebases can be done in a fraction of the traditional time.
  • Keep all documentation up-to-date with AI assistance. Unreliable documentation can be worse than having none; ensure your documentation reflects the latest state of your software.
  • Engage in frequent and detailed conversations about functionality and edge cases. This can help preempt issues and drive better outputs from your AI coding system.
  • Evaluate success not just by how fast you can build, but by how accurately you can define what to build.