All abstractions are wrong, but some are useful.

Analysis of 14 production agent workflows reveals the optimal balance between deterministic code & LLM inference. 65% of nodes run as pure code, saving tokens & improving reliability.

To provide capabilities the platform needs context

Why are people drawn to the Iceberg?

Both the costs to each party and the total costs to the business

How you frame them depends on your audience

It can't all be done by data engineers

It's not just what the spreadsheet says

Pipelines can only be as reliable as their inputs

Only put in fields that drive automation

That's true for tooling and AI agents

Every vendor says their AI analyst works, and the benchmark numbers are high. So I built one in a controlled lab and added grounding one layer at a time (semantic layer, verified examples, knowledge…

How to manage changes of *meaning* as well as structure

SLOs that make a difference

Keep your data contracts lean and useful

A data contract without impact is just documentation.

Automate them instead

Focus on their developer experience, not your own

Look at yourselves too

We all know that when the good people at Gartner highlight a topic as a strategic technology trend for the coming year, everyone will pay more attention – especially within the enterprise ecosystem.…

Or watch them quietly build their own

Simple concept, but powerful in practice

Why Certainty Needs Constraints

Monitoring code quality is as important as data quality

In his classic Programming Perl — affectionately known to a generation of technologists as "the Camel Book" — Larry Wall famously wrote of the three virtues of a programmer as laziness, impatience,…

How do you find the sweet spot?

What if there was a different way to build a data platform, one that is made up of consistent, interoperable, governed data products?