Credible vs. Lightdash

Lightdash is open-source BI on top of dbt — core features MIT-licensed and self-hostable, with agents that answer from its context layer. Credible opens the layer underneath: Malloy as a language, Malloy Publisher as the server, served to every surface rather than to one BI product.

How Lightdash describes itself

What Lightdash is

Lightdash is open-source agentic BI for dbt teams: metrics, dimensions, joins, caching and access rules defined as YAML in your dbt project, served to a metrics catalog, dashboards, embedded data apps and AI agents that answer in Slack, in Teams and from a chart. Its core BI features are MIT-licensed for teams happy to self-host, with a hosted Cloud tier and an enterprise tier carrying advanced security, support and customization.

Category

Credible: AI-native analytics engine with an integrated data stack
Lightdash: Open-source BI for dbt teams, positioned as agentic BI

Who it is built for

Credible: AI product and analytics teams, plus anyone who works with data — spreadsheet users through ML engineers
Lightdash: Analytics engineers and data teams who already work in dbt

Primary interface

Credible: The agent you already use, over MCP — plus Workspaces, notebooks, reports and data apps for the people who want a UI. All of them consume the same model rather than being the place it lives
Lightdash: Its own UI — a metrics catalog, charts and dashboards — with AI agents in the app, in Slack and in Teams

Modeling language

Credible: Malloy — a modeling and query language with imports, inheritance, and public and private members
Lightdash: YAML — metrics and dimensions defined in your dbt project, or in Lightdash YAML if you do not have one

What is open

Credible: Open core. Malloy is open source, and we build and maintain Malloy Publisher, the open-source server for Malloy models. Your model is code in your repository, and Credible is in the Apache Ossie ecosystem for semantic interchange
Lightdash: Open core — core BI features are MIT-licensed and self-hostable; advanced security, customization, support and Lightdash Cloud are commercial

How the model gets built

Credible: A coding agent with open-source MCP tools and agent skills, capturing context from where it already lives
Lightdash: Hand-authored YAML beside your dbt models, with agents that propose changes back to the layer for review

Governance

Credible: Governance as code. Access rules are annotations in the model itself — versioned, reviewed and composable like any other code
Lightdash: Project and space permissions plus user attributes for row-level access, and an agent answers as the person asking

Materialization and caching

Credible: One annotation on the source, in the same file as the logic it materializes — no derived-table block, no rollup definitions, no refresh triggers, no orchestration run to schedule. Optional per source: query your own warehouse directly, or hold a source hot in Credible’s in-memory serving layer, which is how a team on Postgres gets fast serving without buying a warehouse to get it
Lightdash: Caching declared in the context layer alongside the metrics; heavier materialization stays upstream in dbt

How an agent finds the right data

Credible: Search. Typed targets — source, dimension, measure, view, even a dimensional value — are matched against an index of the model and come back ranked, so an agent asks for what it needs instead of picking a model and touring it
Lightdash: The agent selects models and metrics out of the semantic layer, with knowledge documents you upload as extra context, and a router sends a question to the team agent scoped to it

Where the model can be used

Credible: One model served to every surface — agents over MCP, plus APIs, an SDK, embedded dashboards, notebooks and HTML data apps. Results carry an interactive UI resource (MCP Apps, the official extension), so a client that supports it renders a real chart or table instead of the model re-narrating rows
Lightdash: The Lightdash UI, Slack and Teams, a REST API, a Python SDK, an SDK for embedding dashboards, agents and data apps, and the Lightdash MCP server

Scale and portability

Credible: Built for globally distributed, high-availability workloads. Warehouse-agnostic on open-source Malloy — BigQuery, Snowflake, Postgres, MySQL, Trino, Presto and DuckDB, which also reads Parquet straight out of object storage including Azure Data Lake — so the model travels
Lightdash: Self-hosted or Lightdash Cloud, over BigQuery, Snowflake, Redshift, Databricks, Postgres, Trino and ClickHouse

What you pay for

Credible: Usage, not seats — unlimited users on every plan, so adding people never changes the bill. Metered per organization on tokens, bytes processed, bytes served and hot storage, starting free. Your own agent’s tokens are never billed, and a query that runs on your own warehouse is not metered for the scan — only for the result it hands back
Lightdash: A flat platform fee for unlimited users on Cloud, nothing if you self-host, with embedded loads metered separately

	Credible	Lightdash
Category	AI-native analytics engine with an integrated data stack	Open-source BI for dbt teams, positioned as agentic BI
Who it is built for	AI product and analytics teams, plus anyone who works with data — spreadsheet users through ML engineers	Analytics engineers and data teams who already work in dbt
Primary interface	The agent you already use, over MCP — plus Workspaces, notebooks, reports and data apps for the people who want a UI. All of them consume the same model rather than being the place it lives	Its own UI — a metrics catalog, charts and dashboards — with AI agents in the app, in Slack and in Teams
Modeling language	Malloy — a modeling and query language with imports, inheritance, and public and private members	YAML — metrics and dimensions defined in your dbt project, or in Lightdash YAML if you do not have one
What is open	Open core. Malloy is open source, and we build and maintain Malloy Publisher, the open-source server for Malloy models. Your model is code in your repository, and Credible is in the Apache Ossie ecosystem for semantic interchange	Open core — core BI features are MIT-licensed and self-hostable; advanced security, customization, support and Lightdash Cloud are commercial
How the model gets built	A coding agent with open-source MCP tools and agent skills, capturing context from where it already lives	Hand-authored YAML beside your dbt models, with agents that propose changes back to the layer for review
Governance	Governance as code. Access rules are annotations in the model itself — versioned, reviewed and composable like any other code	Project and space permissions plus user attributes for row-level access, and an agent answers as the person asking
Materialization and caching	One annotation on the source, in the same file as the logic it materializes — no derived-table block, no rollup definitions, no refresh triggers, no orchestration run to schedule. Optional per source: query your own warehouse directly, or hold a source hot in Credible’s in-memory serving layer, which is how a team on Postgres gets fast serving without buying a warehouse to get it	Caching declared in the context layer alongside the metrics; heavier materialization stays upstream in dbt
How an agent finds the right data	Search. Typed targets — source, dimension, measure, view, even a dimensional value — are matched against an index of the model and come back ranked, so an agent asks for what it needs instead of picking a model and touring it	The agent selects models and metrics out of the semantic layer, with knowledge documents you upload as extra context, and a router sends a question to the team agent scoped to it
Where the model can be used	One model served to every surface — agents over MCP, plus APIs, an SDK, embedded dashboards, notebooks and HTML data apps. Results carry an interactive UI resource (MCP Apps, the official extension), so a client that supports it renders a real chart or table instead of the model re-narrating rows	The Lightdash UI, Slack and Teams, a REST API, a Python SDK, an SDK for embedding dashboards, agents and data apps, and the Lightdash MCP server
Scale and portability	Built for globally distributed, high-availability workloads. Warehouse-agnostic on open-source Malloy — BigQuery, Snowflake, Postgres, MySQL, Trino, Presto and DuckDB, which also reads Parquet straight out of object storage including Azure Data Lake — so the model travels	Self-hosted or Lightdash Cloud, over BigQuery, Snowflake, Redshift, Databricks, Postgres, Trino and ClickHouse
What you pay for	Usage, not seats — unlimited users on every plan, so adding people never changes the bill. Metered per organization on tokens, bytes processed, bytes served and hot storage, starting free. Your own agent’s tokens are never billed, and a query that runs on your own warehouse is not metered for the scan — only for the result it hands back	A flat platform fee for unlimited users on Cloud, nothing if you self-host, with embedded loads metered separately

A different premise

Where Credible takes a different approach

These are differences in what each product set out to be, not faults in Lightdash. Credible was built the other way around from a BI tool: capture what your data means, enrich it into a governed model in open-source Malloy, and serve that meaning over MCP to the agent you already use — with a UI of our own available, but never assumed.

Two things it can mean to be open

Lightdash MIT-licenses its core BI features for teams that want to self-host, with advanced security, support and customization on its commercial tiers, and the metrics stay as YAML in your repository either way. Malloy and Malloy Publisher are MIT as well, so the licence is not what separates the two. What sits under it is: Lightdash opens the BI application, Credible opens the language and the server that runs the model, and the commercial product sits above them. Either way you can host something real — the difference is whether the open piece is the tool you look at or the format your model is written in.

A language, and YAML

YAML beside your dbt models is quick to pick up, easy to diff, and it keeps the metric next to the table it comes from — a large part of why Lightdash fits dbt teams as well as it does. Malloy is a language: imports, inheritance, and public and private members, so a definition is written once and composed rather than repeated. That composition is what keeps a very large model readable to people and safely extensible by an agent.

Two ways for an agent to find a field

A Lightdash agent selects models and metrics out of the semantic layer, with knowledge documents you upload for extra context and a router that hands a question to the team agent scoped to it — which keeps each agent inside a small, well-understood slice of the business. Credible searches instead: typed targets — source, dimension, measure, view, even a dimensional value — matched against an index of the model and returned ranked, so one agent asks for the handful of fields a question needs rather than being scoped to a subject area up front.

Which way round the platform is arranged

Lightdash builds dashboards, agents and data apps on a context layer, and BI is what those add up to. Credible is arranged the other way round: the model is the product, and Workspaces, notebooks, reports, data apps, embeds, APIs and agents over MCP are all consumers of it. You get a BI surface either way; the difference is whether the model exists to serve one, or the surfaces exist to serve the model.

Materializing as one annotation

Lightdash declares caching in the context layer next to the metrics and leaves heavier materialization upstream in dbt — a clean division of labour that keeps the BI layer light and puts performance work where dbt is strongest. Credible puts it in the serving path: one annotation on the source materializes it, in the same file as the logic, with no model file, config block or orchestration run to keep in step, and the result is held hot in our in-memory serving layer rather than waiting on a warehouse scan. It stays optional per source, which is what makes it useful to a team on Postgres that wants a few views served fast without buying a warehouse for the privilege.

Other comparisons

Compare Credible with the rest

Capability claims about Lightdash were checked against public sources in August 2026. Products in this category change quickly — confirm anything decision-critical with the vendor. Sources: Semantic layer, AI agents and the context layer — Lightdash documentation; MIT licensing, self-hosting and warehouse adapters — Lightdash GitHub and docs; Open-source, Cloud and enterprise tiers — Lightdash pricing page; Series A — TechCrunch, tech.eu, October 2024.

Ready when you are

Bring your own data and judge for yourself

See what a governed model looks like when agents, not dashboards, are the primary consumer.