Puttshack USA

Todd Spurlock

Intelligent Delivery Advisor

Building Faster and Smarter Software Delivery

Experience has taught me that frequent small releases are better than “big bang” releases. With small releases, each team member can be given ownership of that feature’s related user stories. They own the development and testing. They decide when it’s ready to release. 

To improve speed, quality and reliability among development lifecycle, we have adopted AI tooling across the entire SDLC process. We use AI assisted design to build our UI. We then use AI to convert that design into actual visual components. For the backend parts of the system AI is used to assist in the crafting of REST api’s as well as debug issues. AI subagents are used to craft unit tests as well as interactive “browser” testing the running final product. It is important to note this automation does not work without careful planning. Knowing where to split AI tasks into subagents and skills is critical to success. 

Balancing Speed with Engineering Discipline

As work comes in we assess the level of effort to insure quality and the amount of risk involved in making the enhancement or behavior change. This feeds into a testing plan for the feature. If the risk is high enough, we will plan to do a phased rollout, slowly releasing the change to our venues over several days or even weeks if necessary. One of the most important lessons I have learned is that, people are at their best when you include and empower them. Involve them as much as possible in decision making. 

 AI has changed the game. It’s no longer important that you know a particular tech stack or programming language. 

Start small and build on it, take calculated risks. Learn and adjust from early failures, feed those lessons into the next attempt, but keep trying. 

To the professionals who are looking to build a career in software engineering and modern development practices, we are at another great transition point. AI has changed the game. It’s no longer important that you know a particular tech stack or programming language. It is now far more important that you can hear a need, and formulate a plan to full fill that need. The age of having others do that for you and tell you what to code are over.