What it is
Lightdash is an open-source “agentic BI” platform for data teams: metrics, dashboards, AI agents and data apps built from a governed context layer, with analytics kept as code (GitHub README, 2026-10-05). It is built around dbt-style YAML definitions. See Open-source BI comparison.
Maker and history
Operated by Telescope Technology Limited, trading as Lightdash (repo LICENSE, copyright 2021-present); GitHub repo created 2021-03-19 (GitHub API). Funding: not verified.
Products and editions
Lightdash Cloud (hosted), self-hosted (Docker/Kubernetes), open-source core; enterprise features need a licence key when self-hosted. Free and paid tiers are mentioned in vendor material but not verified; see https://www.lightdash.com/pricing (not fetched). Licence: the repo LICENSE says everything outside packages/backend/src/ee is MIT; the ee directory has its own licence (LICENSE file read 2026-10-07).
Data connectivity and modelling
Warehouse adapters: BigQuery, Snowflake, Redshift, Databricks, Postgres and others. Semantic/context layer: metrics, dimensions, joins and access rules defined in YAML alongside dbt models; CLI preview, CI validation, pull-request review.
Deployment
Cloud, self-hosted, embedded (SDK for dashboards and agents with row-level security).
AI features as of 2026-10-05
- AI agents: ask questions in plain language, get governed answers, charts, dashboards, data apps; use the semantic layer, respect project permissions and user attributes, return inspectable queries. Generally available for core conversational agents; an add-on available for all plans (docs, re-checked 2026-10-07).
- Beta (docs 2026-10-07): content-editing tools, AI coding agent, issues findings. Autopilot is a separate admin-only scheduled agent for project maintenance.
- Governance: agent tag scoping, verified answers, evaluation suites, issue tracking.
- MCP: Lightdash MCP server (use from Claude, ChatGPT, Cursor), external MCP connections from agents (Notion, Linear, Confluence), and a Docs MCP for coding agents.
Strengths and weaknesses (opinion)
- Strengths: built around dbt-style YAML, so it suits dbt-centric teams; the docs list governance features for AI answers (agent tag scoping, verified answers, evaluation suites). No comparison with other tools was verified.
- Weaknesses: appears to depend on a dbt-style workflow; fewer GitHub stars than Metabase or Superset (about 6.2k stars, GitHub API 2026-10-07, vs 49.5k for Metabase per its note).
Self-learning
- Lightdash docs: https://docs.lightdash.com (free, opened; includes llms.txt index and Slack community).
- AI agents guide: https://docs.lightdash.com/guides/ai-agents (free, opened).
- Lightdash blog: https://www.lightdash.com/blog (opened); “Lightdash University” YouTube channel referenced on the site (channel not opened).
- No vault summaries found in yt-base.
Sources
- https://github.com/lightdash/lightdash (6.2k stars; 6,177 per GitHub API 2026-10-07), https://raw.githubusercontent.com/lightdash/lightdash/main/LICENSE (2026-10-07), https://docs.lightdash.com, https://docs.lightdash.com/guides/ai-agents, https://www.lightdash.com/blog (2026-10-05)
Open items
- Funding, company history, release number not verified; ‘dashboards-as-code’ beta status and ‘agent tag scoping/evaluation suites’ not re-checked.