The open-source BI tool that turns your dbt models into answers - for humans, and now for AI.
Every data team keeps a small, embarrassing secret: somewhere in the company, the same number appears in three dashboards and none of them agree. Lightdash is a business intelligence platform built around a simple, slightly stubborn idea - that the fix is not another dashboard, but a single place where each metric is defined once, in code, and everyone else reads from it.
The company started in 2020 as Hubble, a project in Y Combinator's Summer 2020 batch, before rebranding to Lightdash in 2021. Its founders, Hamzah Chaudhary and Oliver Laslett, had worked together on a data team at a UK insurtech and noticed something that sounds obvious once you say it out loud: the metrics they produced only became useful when they landed inside a BI tool - and no BI tool on the market had actually been designed for the modern data stack they were using.
The modern data stack, in this telling, has a center of gravity called dbt - the tool data teams use to transform raw warehouse tables into clean, tested models. dbt is where the business logic lives. Yet most BI tools treat dbt as an afterthought, re-defining metrics in a separate, proprietary layer that promptly drifts out of sync. Lightdash's answer was to build on top of dbt rather than beside it. Connect a dbt project and Lightdash reads the models, the descriptions, the tags. Define a metric in YAML next to the model that produces it, and that definition becomes the one everyone queries.
This is a modest-sounding technical decision with a large strategic consequence. It puts your metrics in version control. That means you can review them in a pull request, test them in CI, and - the part incumbents quietly dislike - move them somewhere else if you want to. Lightdash likes to describe this as an open semantic layer, and the emphasis on "open" is doing real work. Looker, the tool it is most often compared to, keeps its semantic layer in a proprietary language inside Google's walls. Lightdash keeps yours in your repo.
None of this would matter if only SQL wizards could use it, so Lightdash spends its other half of effort on the people who can't write SQL: the marketer, the operations lead, the founder who just wants to know whether Tuesday was good. They get charts, dashboards, and self-serve exploration without touching a query. The company's own framing is that the product should be "configurable for the SQL experts and intuitive for the rest." That is harder than it looks, and it is roughly the whole game.
Declare metrics in YAML alongside your dbt models. Lightdash auto-generates dimensions and syncs descriptions and tags, so business logic sits right next to the transformations that produce it.
Non-technical users click through governed data to build charts and dashboards. Same definitions, same numbers - no ticket to the data team, no query editor required.
Review, test and merge BI changes with version control, CLI tools, preview environments and CI/CD - so a critical metric never silently breaks in production.
Create scoped agents - a marketing agent, a sales agent - that answer natural-language questions grounded in the semantic layer. An MCP server lets tools like Claude and Cursor query the same governed metrics.
Investors keep backing the same bet: that an open, developer-first BI tool can grow bottom-up through 5,000+ open-source installs and convert into an enterprise business. The October 2024 Series A, led by Accel, brought total funding to roughly $19.4M.
Angels include Shuo Wang (co-founder, Deel) and Michael Grinich (CEO, WorkOS).
Hamzah Chaudhary and Oliver Laslett start the company and join Y Combinator's Summer 2020 batch.
First GitHub commit in April, an open-source BI launch on Hacker News in June, and a rebrand from Hubble to Lightdash.
Raises $8.4M and launches a hosted cloud service with security and governance features.
Accel leads an $11M round with Shopify Ventures and Operator Partners; Lightdash ships its first AI data analyst.
Presents at the Databricks Data + AI Summit, positioning the open semantic layer as the foundation for AI-first analytics.
Lightdash competes with the establishment - Looker, Tableau, Power BI - and with the newer wave of modern-data-stack tools like Metabase, Cube, Sigma and Mode. Its wedge is deliberately narrow: open source, dbt-native, code-first. When you're a small distributed team taking on billion-dollar incumbents, owning one wedge completely beats competing on breadth.
An open-source business intelligence platform built on dbt. It turns your dbt models into a governed semantic layer so teams can explore data, build dashboards and ask AI agents questions - without writing SQL.
Lightdash is open source and dbt-native. Metrics are defined as code alongside dbt models and live in version control, making them portable and reviewable, rather than locked inside a proprietary modeling language.
It was founded in 2020 by Hamzah Chaudhary (CEO) and Oliver Laslett (CTO), originally under the name Hubble as part of Y Combinator's S20 batch.
Roughly $19.4M total, including an $8.4M seed in 2022 and an $11M Series A led by Accel in October 2024, with Shopify Ventures, Operator Partners and Y Combinator participating.
They're designed not to. Agents are grounded in the governed semantic layer, so they answer from defined metrics in your dbt models rather than querying raw tables - reducing hallucinated or inconsistent results.
Product demos and talks: Lightdash on YouTube · search "Lightdash dbt demo"