Don't Go Off Vibes Part 3: What Makes AI Coding Agents Effective

Why do teams using the same AI coding tools see dramatically different results? In the third installment of our Don’t Go Off Vibes series, Stephen Poletto will share new research from Span's latest report analyzing real-world AI coding agent activity. The findings look beyond model selection to uncover the factors associated with lower costs, greater agent autonomy, and less code-review rework. We’ll explore: Why clearer prompts are associated with more cost-efficient AI-generated codeHow codebase and tooling readiness enable agents to work more autonomouslyWhat strong developer oversight looks like—and how it can reduce downstream reworkThe practical changes engineering leaders can make today to improve agent effectiveness Join us for a data-driven discussion about what it really takes to get better results from AI coding agents—and why many of the most important levers are already within your organization. Download the report from Span – with data from over 100 engineering teams – in advance, and come prepared with questions. This will be an interactive event!