AI-augmented analytics engineering at DocuSign: from req to release-ready with Fusion, dbt MCP, and dbt agent skills

AI can already write SQL. The harder question is whether you can trust what it ships. At Docusign, we've built an AI-augmented development workflow on Fusion, the dbt MCP server, dbt agent skills, and Claude Code, one where the agent understands the project it's working in, not just the file it's editing.

In this session, I'll walk through real examples: developing and testing models, modernizing legacy SQL, and running impact analysis before a change lands. Then the part that matters more: the engineering discipline that keeps AI-generated changes production-ready. Repository conventions, agent skills, human review, CI, and deployment controls.

You'll leave knowing where these workflows are delivering measurable productivity gains, where human judgment is still non-negotiable, and how to start adopting the patterns in a mature dbt project.