LoginSign Up
← Back to all sparks
L

Lightdash

ANALYTICS
Velocity6.3

Lightdash is turning the analyst's prompt into the primary way to build BI

business-intelligenceai-nativedata-appsmcpdata-viz
◆Current state
Lightdash is pushing hard on AI-native BI. Its data apps now generate reusable chart types from a plain-language prompt, verified content has gone GA and merged with the AI-agent and MCP layer, and AI-written summaries are appearing in scheduled deliveries. Alongside that, steady core work continues on SQL parameters, chart layouts, and enterprise controls like user impersonation.
◆Where it's heading
The clear direction is a prompt-driven analytics surface backed by a trusted-content layer that external agents like Claude and Cursor can query through MCP. Expect the 'describe it and Lightdash builds it' pattern to spread from chart types into more of the modeling and dashboard workflow, with verification as the guardrail that keeps agent answers trustworthy.
◆Prediction
The next moves likely push prompt-to-artifact generation deeper into dashboards and the semantic model, and expand what the MCP and verified-content layer exposes to external agents.

◆Recent moves

  1. 1d ago

    🧩 Describe a chart type, Lightdash builds it

    ⚡ SPARK

    This extends Lightdash's AI-native push from generating answers to generating the visualization primitives themselves: a described chart becomes a reusable type that behaves like a built-in, usable across explores and dashboards. It is the sharpest expression yet of the prompt-driven direction.

    View source ↗
  2. 2d ago

    ✅ Verified content is now generally available

    Verification graduates from beta and merges with the AI-agent layer, so human-verified and agent-verified content share one store the MCP and external agents can read. It reinforces Lightdash's positioning as the trusted-content source that keeps AI answers reliable.

    View source ↗
  3. 3d ago

    💅 Scheduled deliveries: new look, smarter messages

    Scheduled deliveries get a redesign plus optional AI-written summaries that reflect the actual data sent, generated fresh each send. Another incremental spot where the AI-agents add-on surfaces in day-to-day workflows.

    View source ↗
  4. 15d ago

    🔍 Date zoom, now in SQL

    The date-zoom granularity chosen on a dashboard now resolves inside custom SQL via a reserved parameter, closing a gap between the visual controls and hand-written queries with nothing to wire up.

    View source ↗
  5. 28d ago

    🔀 Reshape your Sankey chart with new layouts

    Sankey charts gain multi-step, merged, and direct node layouts, extending the chart library's flexibility for funnel and flow analysis. Steady breadth work on the core visualization set.

    View source ↗
  6. 1mo ago

    🎭 User Impersonation

    Admins can now impersonate a user for up to 15 minutes, audit-logged, to debug permission and data-access issues — an enterprise troubleshooting affordance rather than an analytics feature.

    View source ↗