LoginSign Up
← Back to all sparks
L

Lightdash

ANALYTICS
Velocity7.5

Lightdash is cutting its dbt dependency and building AI-powered authoring into every layer of its BI stack.

semantic-layerdbt-independenceai-bicustom-chartsgit-backedworkflow-integration
◆Current state
Lightdash is running two parallel expansion tracks: making itself a standalone semantic-layer platform independent of dbt (native YAML with GitHub/Bitbucket sync and AI write-back), and embedding AI throughout the BI workflow — custom chart type generation, deep research, and AI findings that automatically open tickets in Linear and Jira. UX polish releases (URL slugs, sidebar Explorer, per-delivery filters) show a product that has moved past early roughness and is hardening for broader adoption.
◆Where it's heading
The dbt decoupling is the larger structural bet — native Lightdash YAML backed by git repositions the product as a standalone BI and semantic layer rather than a dbt visualization front-end. The AI features follow the same thesis: Lightdash wants findings and model changes to produce actionable outputs (tickets, PRs) rather than just charts. The custom chart type capability, if used broadly, could evolve into a visualization plugin ecosystem. The short-term pattern suggests continued write-back integrations and expansion of the non-dbt path.
◆Prediction
Further write-back integrations are likely — pushing AI findings and semantic layer changes back to more operational tools — alongside continued investment in the native YAML path. Custom chart types, if the generation quality holds, could become a moat; expect Lightdash to expose that surface to a wider set of contributors.

◆Recent moves

  1. 1d ago

    Test warehouse connectivity without deploying

    Splitting 'Test connection' from 'Test & deploy' gives teams a fast connectivity check before committing a warehouse configuration change — reduces roundtrips when debugging a firewall rule or credential typo. Fits Lightdash's broader push to make the platform easier to operate without full deploy cycles.

    View source ↗
  2. 2d ago

    💬 A comments panel for your dashboards

    A comments panel in the dashboard header consolidates all threads in one view, reducing the need to navigate individual chart comments during review workflows. A collaboration UX improvement consistent with Lightdash's ongoing polish cycle.

    View source ↗
  3. 6d ago

    🧩 Build your own chart types

    ⚡ SPARK

    AI-generated custom chart types extend Lightdash from a fixed-chart BI tool into an extensible visualization platform — a structural capability that most tools don't offer. The described use cases (sortable tables, KPI trees, product affinity maps) are real analyst needs that previously required custom code or a different tool entirely.

    View source ↗
  4. 6d ago

    Per-delivery filter overrides for scheduled charts

    Per-delivery filter overrides for scheduled charts eliminate the need to duplicate a chart to send different filtered views to different audiences or contexts. A practical fix for teams running multi-segment reporting from shared chart definitions.

    View source ↗
  5. 6d ago

    ⚡️ GitHub & Bitbucket support for native Lightdash YAML

    ⚡ SPARK

    Native Lightdash YAML now connects directly to GitHub and Bitbucket — with branch-refresh, AI-assisted write-back as PRs, and model management without dbt — making dbt an optional rather than required dependency. This is the most significant structural repositioning in Lightdash's recent history.

    View source ↗
  6. 7d ago

    Chart config sidebar in Explorer removes mode-switching

    The Explorer's new sidebar chart configuration eliminates the mode switch between query editing and chart configuration — a real workflow friction point that made iterative analysis slower. Consistent with the series of UX maturation changes across recent Lightdash releases.

    View source ↗