Most teams finish a project and immediately start the next one. The lessons evaporate. The same mistakes repeat.

Structured reflection is the thing that breaks that cycle. AI just made it faster.

The framework we use

For years, my colleague Lynn Kreun and I have run retros together across government and education — complex, high-stakes environments where the cost of repeating mistakes is high and the appetite for slowing down is low. We call it failing forward.

The framework is simple: Start. Stop. Continue.

• What do you start doing that you weren’t?
• What do you stop doing that isn’t working?
• What do you continue because it’s actually moving things forward?

What changed with AI

What’s changed is that GenAI tools like Kiro and Amazon Bedrock now make this reflection faster, deeper, and more actionable — surfacing patterns across projects that used to require hours of manual synthesis.

The best teams don’t just deliver. They reflect. Then they deliver better.

What’s one thing your team stopped doing after a retro that actually made a difference?

We wrote up the full framework and exactly how AI fits in. Read the full piece on the AWS Public Sector Blog.