AI can be a polarizing topic among technology professionals. Some have dramatically accelerated their work, while others are stuck with mediocre results that leave much to be desired. So what separates these outcomes? How are some teams achieving 10x acceleration with impressive accuracy while others can't get an AI project off the ground without completely rewriting everything the AI produced?

This session follows two AI implementation journeys. One builds a program on best practices that enable developers to use AI as a powerful pair programmer. The other is an experiment in futility where vibe coding is allowed to run unchecked. The difference? One has a foundation built on knowledge management, context engineering, and well-crafted user stories. The other thrusts large language models into the hands of developers with no training on how to work with AI as a partner.

Collin Newberry, Director of DevSecOps at Yahara Software and author of The DevOps Problem, will demonstrate what separates successful AI implementations from expensive disappointments and show the contrast between structured pair programming and undisciplined vibe coding in real time.
In this session you will learn:

- The foundational components every successful AI implementation depends on
- How context engineering sets AI agents up for success
- Why prompt engineering remains a critical developer skill
- How well-written user stories help AI produce better code
- The measurable difference between AI pair programming and vibe coding