← Back to all sparks
L

Lightdash

ANALYTICS
Velocity7.5

Lightdash ships AI-described custom chart types and a content governance overhaul in one week

analytics-platformcustom-chartscontent-governancedbt-nativeenterprise-biai-integration
◆Current state
Lightdash is shipping on two simultaneous tracks: platform extensibility (custom chart types, nested column support for BigQuery/Databricks, native YAML without dbt) and enterprise governance (verified content promotion workflows, dashboard ownership, duplicate detection, role-based edit locks). The custom chart types feature is the most notable — users describe a chart in natural language and Lightdash builds it into a reusable project-level chart type, extending the BI tool beyond its fixed chart library.
◆Where it's heading
Lightdash is converging on an enterprise-grade BI platform from its dbt-native analytics tool origins. Supporting native YAML without dbt, enabling custom chart types, and building content governance infrastructure all point toward a product that no longer requires dbt as a prerequisite and competes more directly with Metabase, Looker, and Tableau for the data team segment. The AI agent integration (findings pushed to Linear/Jira) is early infrastructure for a proactive monitoring layer.
◆Prediction
The AI agent path that pushes findings to issue trackers is likely to deepen into scheduled anomaly detection and threshold-based alerting. Proactive BI — where the tool surfaces what changed without anyone asking — is the natural evolution for a platform that already has AI agents and scheduler infrastructure in place.

◆Recent moves

  1. 3d ago

    🪆 Explore nested and repeated columns

    Lightdash adds support for exploring nested and repeated columns on BigQuery and Databricks natively, referenced by full path in YAML without flattening. Arrays are queried one row per element with automatic warnings when mixing multiple arrays could inflate metrics. This closes a meaningful gap for teams working with nested data models who previously had to pre-flatten their schemas.

    View source ↗
  2. 9d ago

    ✅ Promote, own, and find verified content

    Verified content gets a full governance workflow: submission queue, reviewer notifications, duplicate flagging at submission time, dashboard ownership assignment (user or group), verified status filtering in homepage and search, and a new custom role scope that restricts editing of verified content. This is the most substantial governance update Lightdash has shipped — moving from a simple verified badge to a full content promotion and ownership model.

    View source ↗
  3. 11d ago

    🔌 Test your warehouse connection before you save it

    A new connection test option checks reachability without compiling the project — useful for validating credentials and firewall rules before committing warehouse connection changes. DX improvement with no new functionality.

    View source ↗
  4. 12d ago

    💬 A comments panel for your dashboards

    A new Comments panel in the dashboard header surfaces all comment threads in one place, removing the need to open individual tiles to read discussions. Minor UX improvement.

    View source ↗
  5. 16d ago

    🧩 Build your own chart types

    ⚡ SPARK

    Lightdash introduces extensible chart types: users describe a chart in natural language and Lightdash builds it into a reusable chart type for the project. Anyone on the team can then select it, map their fields, and save it like a built-in chart. This changes Lightdash from a tool with a fixed chart library into a platform where analytics patterns — KPI trees, product affinity matrices, interactive tables — can be institutionalized as project-level primitives.

    View source ↗
  6. 16d ago

    🎯 Different filters for each chart delivery

    Scheduled chart deliveries can now carry per-schedule filter overrides — the same saved chart can be delivered to different stakeholders with different data slices. Useful for regional report distributions or department-specific views from a single chart definition.

    View source ↗