What Data Agent Benchmarks Do and Don't Tell Us
The recent AI Council conference highlighted a universal pivot toward AI infrastructure, with companies ranging from startups to established players redefining themselves as providers of context, orchestration, or compute layers for agents. New AI‑native databases such as LanceDB are being built from the ground up to serve LLM‑driven workloads, signaling a shift from retrofitted analytics stores. Benchmarking efforts like ADE‑bench reveal that agents excel on well‑specified tasks, yet current tests overlook stateful, long‑term performance and the value of rich organizational context. Attendees emphasized that token‑efficient, context‑aware agent workflows will be critical as usage scales.
Moving Up the Stack: Analytics Engineering in the Age of Agents
The article argues that analytics engineering must “move up the stack” again, this time leveraging AI agents to automate routine data work. It highlights dbt’s meteoric growth—over three million daily downloads and a billion total downloads—showing how the tool already reshaped...
Agent Skills: Disseminating Expertise
dbt Labs unveiled a suite of eight AI agent skills that automate complex dbt tasks, including a migration from dbt Core 1.10 to Fusion that completed without human intervention. These skills distill hundreds of hours of community expertise into concise...
The Iceberg Ecosystem Today (Anders Swanson)
The data industry is rapidly converging on open standards, and dbt Labs is leading the charge by migrating its entire data stack to an Iceberg‑based lake that supports multiple compute engines. In a recent podcast, Anders Swanson outlined the current...
AI Agents and the Data Lake (W/ Lauren Anderson)
In this episode, Tristan Handy talks with Lauren Anderson, head of Okta's enterprise data platform, about how identity underpins the emerging challenges of AI agents and open data lakes. Lauren explains the need for central governance and a shared semantic...