Changelog
The product moves every week.
Everything that shipped in BeryX — new features, improvements and security hardening — generated straight from our delivery cycle.
19 updates · last updated September 23, 2026
Week of
- Improved
Charts in exported reports
HTML report exports now include charts.
- Improved
More technical depth in reports
Reports include remediation in SQL, pandas and PySpark, plus context for medallion, data mesh and ML architectures.
- New
Incidents
Issues related through data lineage are grouped into a single incident, so you fix the cause rather than the symptoms.
- New
Automatic monitors
Freshness, volume and schema changes are monitored by default on every dataset, with no rules to configure.
- Improved
Deeper profiling
Quantiles, histograms, correlations and automatic alerts, computed directly in your database.
- Improved
Feedback on AI outputs
Rate AI reports, analyses and suggestions with 👍/👎 — your feedback improves future versions.
- New
Knowledge graph
An interactive view linking your workspace's sources, datasets and knowledge.
- New
Knowledge page
Search and curate your workspace's knowledge base on a dedicated page.
- New
Knowledge base
Reports and root-cause analyses now feed a knowledge base the assistant draws on when answering.
- New
Issue kanban board
Organise issues on a kanban board with drag and drop, filters and due dates.
- New
Issue lifecycle
Data-quality issues now have states, a full history and an assignee.
- New
Workspace report by persona
A workspace health report tailored to engineers, managers or executives.
- Security
Safer sessions
Resetting your password signs out every open session, and accounts now require a verified email.
- Security
Customer-only credentials for GCS, S3 and BigQuery
GCS, S3 and BigQuery connectors only use the credentials you configured — never the server's own.
- New
Scheduled executive reports
Schedule executive reports per workspace and receive them by email at the cadence you choose.
- New
MCP server over HTTP
DataVault exposes an MCP server over Streamable HTTP, with resources, so AI agents can use your data-quality context.
- Improved
Document scan status in the UI
Queued document scans now show their status in the interface, from request to completion.
- New
Rules generated from profiling
Profiling each dataset now proposes deterministic quality rules you can enable without writing SQL.
- New
Ask DataVault
Chat with your data in plain language: the assistant answers from your workspace's scans, rules and incidents.
Want to see this on your own data?
Create a free workspace or talk to us about a diagnostic.