Beyond Pipelines
Agentic Data Engineering
Data engineering is being rewritten. Agents can now discover data, check for existing solutions, assemble dataflows, and test changes—but only if the underlying system gives them full context, native lineage, versioned tables, and safe human gates. Without that substrate, agents hallucinate or break things. With it, data requests that used to take months can resolve in hours.
This publication explores what that shift actually requires. We write from the operator’s side of the problem: what we learned building data systems through the last major transition, what still fails when agents are bolted onto disconnected tools, and the architectural properties that make agentic data engineering reliable.
Who it’s for
Data leaders, platform engineers, and data engineers responsible for backlogs, AI readiness, and production reliability.
What you’ll get
One substantive essay every two weeks. Clear arguments, concrete patterns, and lessons from building systems that have to work under real constraints. Occasional technical deep dives with Alejandro Abdelnur (Tucu), co-founder and CTO.
We are building Tabsdata, an agentic data engineering platform that treats the agent and the infrastructure as one system. The writing here is about the category and the requirements; the product is one implementation of those ideas.
Subscribe if you want the thinking, not the pitch.Why subscribe?
Subscribe to get full access to the newsletter and publication archives.
