✨ We're thrilled to introduce Ella — our new AI layer for data observability! ✨ Ella includes 5 different AI agents that will help you with every task in your workflow: • ✅ Test Recommendations Agent — Recommends and writes the right tests—based on your current coverage policy, best practices, lineage, and what’s missing. • 🧠 Governance Agent — Flags gaps in ownership, documentation, and tagging—and generates fixes to help make your data AI-ready. • 🛠️ Triage & Resolution Agent — Pinpoints root cause and opens a PR with the resolution. • 📚 Catalog Agent — Makes metadata searchable and useful for both users and AI tools. • 💰 Performance & Cost Agent — Finds performance issues and recommends what to rewrite or retire. These agents are interactive. They take context, iterate with you, and complete observability tasks autonomously. ⚠️ This is an opt-in beta. If you’d like early access, DM us directly. 🔐 Security note: agents use hosted models only, do not extract sensitive data, and follow strict AI governance policies. 📝 Activation requires signing our AI policy agreement. Read the full announcement.

<!here> If you're with us on the Ella webinar, feel free to ask questions here!

Will the AI agent will be able to understand the context of the tables and column without the descriptions? Asking because documenting the tables and column is one of the main challenges we have

is you catalog compatible with other solution for the market or can replace them ? What if in my organisation not all my flows are covered by DBT ( meaning not visible by elementary) ?

sj.7272 from Project72 was added to this channel by noya. You can review their permissions in Channel Details. Happy collaborating!

Will the AI agent only have access to metadata and/or is there any place specific in the documentation to point to that for something like a companies AI Council to review?

Enhanced Alert Management🚨 Easily see where alerts go and fine-tune delivery in one place📍 The new Alert Destinations tab is now available in both the test configuration page and the test overview side panel. • See exactly where alerts are being sent for each test — quickly identify any gaps in your alerting strategy. • Bulk enable or disable alerts across multiple tests for faster setup and adjustments. • Easily override routing settings with just a few clicks (coming soon). Read more in our docs. Resolved Alerts Now Sent in the Same Thread 🧵 When an issue is resolved, the alert is now sent as a reply in the original alert thread—instead of a new message. Owners and subscribers are tagged so nothing is missed. Disable Alert Prioritization 🚦 Out of beta and ready for use—alerts can now be sent to all applicable rules. Previously, alerts were sent based on the first matching rule. Now, you can choose to have alerts delivered to all matching rules instead. This gives you more flexibility and control over alert routing across teams, domains, and use cases. DM us if you’d like this enabled in your workspace. Catalog Improvements 📚 Test Coverage Visibility in the Catalog 👀 Instantly see where your data is covered — and where it’s not. We’ve added a powerful new Test Coverage View to the Catalog, allowing you to take control of your data quality strategy with confidence. Now you can: • Visualize test coverage by dimension — freshness, completeness, uniqueness, validity, accuracy, and consistency. • Identify gaps and uncover datasets that require further attention. • Add missing tests directly from the Catalog in just a few clicks. Read more in our docs. Add Tests Directly from the Catalog 🛠️ You can now add tests in two new ways—right where you need them: • From the Columns tab to add column-level tests • From the Dependencies tab to add tests to upstream/downstream tables You can also see the latest test results per column and table to spot gaps and take action, without switching tabs or tools. Add Governance Metadata to Your Entire Pipeline — in Seconds 🏷️ From the Dependencies tab in the Catalog, you can now effortlessly add governance metadata — like tags, owners, and descriptions — to all upstream and downstream assets in one go. No more hopping between tables. Now you can scale your metadata coverage and keep your pipeline well-documented with minimal effort. Automated Monitors Improvements🧪 Understand Why Automated Monitors Show “No Data”💡 Freshness and volume monitors now explain why a “No data” status appears, so you can troubleshoot faster. You’ll see specific reasons like: • Not enough data during the training period • Missing permissions • No baseline pattern found • Table hasn’t been updated recently This helps you quickly understand what’s happening and decide what (if anything) needs to be fixed. Hour-Level Detection 🕙 You can now set detection periods in hours for automated monitors, not just days. Previously, monitors could only detect issues on a daily or longer basis. Now you get higher-resolution visibility into your data, with the ability to detect issues closer to when they actually happen. This is especially useful for datasets that update multiple times a day or require faster feedback loops. Zoom Into Your Anomalies Graphs 🔍 You can now drill down into anomaly results with higher-resolution data—so you can see exactly what changed, when it changed, and how it compares to expected behavior. It’s a clearer, more detailed view that helps you investigate with confidence. User Activity Logs 📝 Admin roles on the platform can now export logs of recent user actions, like adding, modifying, or deleting tests, monitors, metadata, and more. It’s an easy way to stay on top of changes, support audits, and ensure everything’s running smoothly. DM us if you’d like this enabled in your workspace. Learn more on our docs. Support for groups Metadata from dbt 👥 Elementary now collects the

groups

field from your dbt models. If you use the

groups

metadata field in dbt, it will now be included in Elementary’s metadata schema. You can use this field to organize models, filter assets, and structure ownership or domain views across your catalog and coverage dashboards. It’s one more way to align your dbt structure with how your team works. New Integration - Clickhouse ⚡ You can now monitor and test your ClickHouse pipelines with Elementary. Run data tests, track freshness, and collect metadata across your ClickHouse models—just like you do with Snowflake, BigQuery, Redshift, and other supported warehouses. Read more in our docs.

Project72 has removed themselves from this channel.