When skills meet scale: an agent assembly line for dbt

At dbt Summit 2026, Riot Games senior data engineer Jess Zhang described using AI agents and dbt to rebuild a production data warehouse for the Echoes on the Rift event, which required analysis of billions of game events and more than 60 cross-title metrics. The approach addressed the complexity, inconsistent testing, and limited accessibility of legacy PySpark workflows through a layered context system combining engineering-defined agent instructions, shared analyst and engineer playbooks, and simple analyst task prompts. Its five-stage agent assembly line discovers and profiles source data, generates silver and gold models, and validates SQL through syntax, compilation, and execution checks before human pull-request review. Although the event project took roughly two and a half months compared with two months for a similar manually built project, it was completed alongside on-call work, changing requirements, and development of the agent system itself. Zhang argued that the primary benefit is a lower marginal cost for additional models after investing in reusable playbooks and evaluation systems, while engineers remain responsible for data-modeling decisions, business semantics, and review; avoiding redundant work across contributors remains an unresolved challenge.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.