Blog
What it takes to hand AI the work
Notes on data and AI in operations: what it takes for AI to earn a workflow, from checked answers to AI-run operations.
-
Why LLMs fail on querying enterprise data
Why LLMs fail at querying enterprise schemas: because the business rules were never in the schema.
-
How we're building a growth engine that runs itself
How we're building an engine that runs a growth cycle end to end, leaving only the decisions to the team.
-
Your semantic layer is currently a person
What a semantic layer is, why yours is probably a person, and why AI turned it into the thing that decides whether a number is worth acting on.
-
Evolving the enterprise data stack for AI
Why enterprises don't trust AI with business decisions yet, and the three changes to the data stack that fix it. Delete, derive, reinvent.