The Rise of the Agentic Data Catalog
Prukalpa, Mastercard's Zhenni Hu, and leaders from ASOS and Sophos on what shipping AI-generated context at scale took. Recorded September 10, 2026.
What you'll see
01
Why catalogs kept failing, and what closes the loop
Prukalpa traces a decade of catalog failure to three causes — manual supply, siloed consumption, and businesses that outrun their documentation — and shows how context agents, conversational search, and a learning engine answer each one.
02
Mastercard's three decisions
Zhenni Hu held back from running AI across everything. A bounded set of critical assets, drafts reasoned from real usage and lineage, and an expert sign-off before anything goes live: 30,000 assets enriched in two weeks, with roughly half needing no edits at all.
03
How ASOS and Sophos scaled it
Sophos enriched 43,000 assets, then built a harness that graded every AI description A to E before the business saw it. ASOS sequenced left to right along the pipeline so each layer of context compounds the next.
04
The steward's job moves up
Less drafting, more judging. All three teams describe governance shifting from a compliance mandate to a function other teams seek out, with agents becoming the catalog's newest consumers.
Explore the context layer with Atlan
WTF is the Context Layer?
A bi-weekly live series for AI leaders. One burning question per episode, an open AMA, and guests who've actually built context layers.
Register for the Series
The Enterprise Context Layer
Guides on what the context layer is, how it works, and how teams are building it for AI agents on Snowflake, Databricks, and dbt.
Explore the Guide
See the Context Layer in Action
Watch how Atlan bootstraps enterprise context from your existing data estate, standardizes it, and deploys an AI analyst.
Watch Demos