AI-augmented analytics engineering at DocuSign: from prompt to production with dbt v2, dbt MCP, and agent skills
DocuSign wired Claude Code, dbt Agent Skills, dbt MCP, and dbt v2 into one workflow and measured a 50% productivity gain.
Overview
Explores DocuSign's AI-assisted data development—model development, SQL modernization, impact analysis—while maintaining production quality. Emphasizes that trusting AI output requires repository conventions, agent skills, human review, CI pipelines, and deployment controls.
What you'll take away
- 01
SQL generation was never really the bottleneck at DocuSign; context switching between a Jira requirement, unwritten dbt conventions, and repeated validation was.
- 02
Splitting the workflow into discovery, dbt development, and orchestration phases meant a failed compile or skipped test gets retried on its own instead of restarting the whole pipeline.
- 03
The pilot delivered a 50% productivity gain from requirement to a validated, review-ready dbt asset, with human review still required before anything merges.
Session speaker
Bishal Gupta
Docusign