/ Feature
Automated Estate Scanning
Oracle, SQL Server, Snowflake, S3, Excel and Postgres scanned directly — so coverage follows the estate as it changes rather than trailing months behind whoever last had time to document it.
Solution · GenMeta
A catalogue that writes itself.
Enterprise data estates span dozens of disconnected source systems with no consistent metadata, ownership or lineage — and manual cataloguing cannot keep pace with schema drift, undocumented columns and personal data quietly accumulating where nobody expects it. GenMeta scans the estate itself, describes what it finds, and keeps scoring it as things change.
01
/ Challenges addressed
Why catalogues go stale.
- 01Dozens of source systems with no consistent metadata, ownership or lineage
- 02Manual cataloguing that cannot keep pace with an estate that keeps moving
- 03Schema drift arriving unannounced, invalidating documentation nobody reread
- 04Undocumented columns that only the person who built them can explain
- 05Personal data accumulating in places no architecture diagram records
/ What it does
Scanned, described, scored.
/ Feature
AI-Generated Descriptions
Column-level descriptions generated from the data and its context, giving every field a starting explanation instead of a blank entry nobody will ever fill in.
/ Feature
Automatic PII Labelling
Personal data identified and labelled wherever it turns up, including the copies, debug tables and free-text fields that never appear on an architecture diagram.
/ Feature
Continuous Quality Scoring
Data quality scored on an ongoing basis rather than assessed at project milestones, with drift and quality issues surfaced in the dashboard as they appear.
/ Feature
Knowledge Graph
Relationships between datasets, columns and concepts held as a graph, so questions about how things connect can be traversed rather than guessed at.
/ Feature
Ask GenMeta
A natural-language interface over the catalogue — ask where something lives or what a field means in plain English, rather than knowing the official dataset name in advance.
/ Why we built it
Automate the transcription, keep the judgement.
01
Harvest before you curate
Schemas, volumes, null rates and lineage can all be observed. Asking a person to type them is why most catalogues stall at five per cent coverage.
02
Suggest, then confirm
Classification and relationships are inferred and put in front of a person to accept. Confirming a suggestion takes seconds; writing one from scratch takes minutes.
03
Find what nobody declared
The valuable answer is not where personal data is supposed to be. It is where scanning finds it, in the copies and columns nobody remembered.
/ Let's talk
We build platforms like this.
GenMeta is one of several solutions we've built in-house. If you cannot answer where your sensitive data lives, we'd be glad to talk about it.