Your warehouse
- Amazon Redshift
- BigQuery
- ClickHouse
- Databricks
- MotherDuck
- PostgreSQL
- Snowflake
- Trino
Source of truth
Platform
Lightdash Context Layer
Semantics
Metrics · joins
Context
Business logic · internal docs
Caching
Fast · consistent results
Permissions
Authentication
Activation
Consumption layer
Agents
Governed autonomous AI
Apps
Data Products · Workflows
Dashboards
Interactive · drill-down
SDKs
Python · JS · API
Embedding
Customer-facing analytics
MCP
Internal / external
Evaluation
Feedback loop
Autopilot
Scheduled health, zero effort
Self-improving SL
Metrics that never drift
Caching
S3 · Pre Aggs
Context
.md · .files
Permissions
RLS · Auth · SSO
Sends to context layer
Continuously refined
The context layer behind data teams at:
Why use the Context Layer?
Yes - one definition per metric in code
Partial - trapped in the BI tool
No - guesses from raw tables
Yes, every answer
Rarely
No
Yes, in the context layer
Varies, tool-specific
No
Yes
Limited
No
Yes (dbt/Lightdash YAML)
No
No
Yes
Varies
No
Yes
Rarely
No
“If you’re tired of fragmented reporting, duplicated work, and inconsistent numbers, Lightdash is the fastest way to get to a single source of truth."
Matthew Powers
Chief Technology & Product Officer @ Tatango
40%
reduction in data team hours
100%
alignment on key metrics
50%
faster to create dashboards
For any other questions, reach out to our team.
What is the Lightdash context layer?
It's the governed layer Lightdash runs on top of your warehouse. It holds your metric definitions, business context, permissions, and caching in one place. Every agent, dashboard, app, and SDK reads from it, so everyone (and every LLM) works from the same source of truth.
Why not just point ChatGPT or Claude at our warehouse?
A raw LLM has no governed metrics, no permissions, and no business context, so it guesses and can return confidently-wrong numbers. Lightdash agents answer only through the context layer, querying your defined metrics. Every answer traces back to its official definition, so you get accuracy you can audit, not fiction.
How does it stop AI from making up metrics?
Agents can't query raw tables freely. They answer through your defined metrics in the Context Layer. The richer your descriptions and business context, the more accurate the answers. Every response traces back to the metric and query behind it, so anyone can check it.
How do permissions and security work for enterprise?
Permissions live in the Context Layer, including row- and column-level security, with SSO and access derived from your identity provider. The same rules apply to dashboards, apps, embeds, and AI agents. Lightdash is SOC 2 compliant, with self-hosted and private/on-prem deployment options.
Do I need dbt?
No. Most teams build their Context Layer inside an existing dbt project, but you can define it directly in Lightdash YAML against your warehouse tables with no dbt at all. The syntax matches dbt's, so adopting dbt later is straightforward.
What does it cost?
Lightdash is flat-rate with unlimited users — no per-seat pricing. You can open governed access to your whole company. Enterprise plans add private deployment, advanced security, and SSO. Book a demo for specifics on your setup.
I have an infrequently asked question.
Great! We're always around to help. You can chat to us through the live chat on this webpage, or talk to us in our Slack community.
