What is ADL?

ADL activates your design metadata. Capture how your data solution is structured — the objects, items, mappings, classifications, and relationships — and turn it into whatever your team needs: code, documentation, compliance views, deployment scripts, tests. If it's text-based, ADL can generate it.

Most teams have a design somewhere — in diagrams, modeling tools, slide decks, or someone's head. ADL takes that design, captures it as structured metadata in your own repo, and connects it to templates that produce everything downstream. Engineers see the code that builds the warehouse; modelers see relationships and business keys; architects see methodology and layers; stakeholders see compliance and capability views. Same metadata. Different activations.

ADL runs in your browser. No database to install, no server to manage, no special tools to set up — just a local folder and a modern browser. Your metadata stays on your own machine, in your own repository.

First — what is design metadata?

Most people hear "metadata" and think of the dry catalog stuff: column types, table sizes, the bookkeeping that database administrators worry about. Design metadata is different. It is the structured expression of how your data solution is meant to work — the conventions, the source-to-target mappings, the data definitions, the transformation rules. It is the design itself, captured as data the rest of your toolchain can read.

An example: when you decide that Customer is a core business concept tied to a specific business key, and that data from three source systems should flow into it via mappings that handle deduplication, the result of those decisions is design metadata. The design lives somewhere — in a diagram, a spreadsheet, someone's head — but it only becomes activatable when it's expressed as structured metadata.

This kind of metadata is the true intellectual property of a data solution. It captures the unique and custom way an organization interprets its data — interpretations that don't exist in any open standard, because they're specific to the business. Off-the-shelf modeling tools store this IP inside their own databases, where it depreciates the moment the tool falls out of favor. ADL keeps it as open-format files in your own repository, where it stays yours.

For a much deeper treatment of this — the principles, the patterns, the philosophy behind treating design as the durable artefact — see Data Engine Thinking by Roelant Vos and Dirk Lerner.

How it works

ADL is built around three simple ideas.

1. Metadata — describe what you have

Your data solution starts with metadata. In ADL, this means defining your data connections (where your data lives), data objects (your tables, views, queries, or any structured thing you want to model), and how they relate to each other through mappings. You can also tag everything with classifications (your taxonomy) and govern it with conventions (your naming and structure rules).

All metadata is stored as plain JSON files using an open-source schema. No proprietary format and no database to manage — just files in your repository that you can edit in ADL or any text editor.

2. Templates — describe what you want

Templates define the output you want to generate. They're written in Handlebars, a straightforward templating language that's easy to pick up even if you've never used it before.

A template might generate a SQL CREATE TABLE statement, a stored procedure, a Markdown documentation page, a deployment script, a test fixture, an API call, a configuration file — anything text-based. ADL ships a library of ready-made templates for common patterns, and you can write your own for anything else.

3. Output — generate and deploy

When you combine metadata with templates, ADL generates your output files — into your project folder, ready to be reviewed, committed, and deployed through your normal CI/CD pipeline. Each template generates one output file per mapped metadata object, so everything stays organized and predictable.

A single data object can be mapped to multiple templates, so the same captured design can simultaneously produce a table script, a stored procedure, a documentation page, and a configuration file. One design, many activations.

Your design, your way

ADL gives you four things to configure: how you structure your design, how you classify it, which perspectives you view it through, and what you generate from it. The framework connects them; the lenses are yours.

ADL ships starter classifications, personas, and templates for common patterns — Data Vault, Persistent Staging, compliance views, conceptual modeling. The framework is the product. If you can describe how your design should be categorized, viewed, and rendered, ADL can carry it.

What ships in the box

Install a starter solution from the Marketplace — you get pre-configured metadata, templates, and conventions. Open the project and start generating immediately.

Beyond the starter solutions, ADL ships an open-source template library covering staging, integration, Data Vault, and documentation patterns, plus a set of custom Handlebars helpers for string operations, lookups, and test-data generation. Replace them, extend them, or write your own.

What we believe

• Your design choices should not be constrained by a tool's default opinion. Methodology, platform, and how you organize your design — those decisions belong to your team, not your vendor.
• Open formats outlast proprietary ones. Metadata and templates are plain files in your own repository — readable, diff-able, portable.
• Speed and autonomy are not opposites. Starter configurations give you a fast start; configurability lets you evolve without replatforming.

More on the About page.

Get started

ADL is free in public preview. The quickest path is to install a starter solution from the Marketplace, run the generator, and see real output land in your repository within minutes.

• Try the app — free, no signup hurdles.
• Quick Start Guide — set up your first project in a few minutes.
• How ADL Works — the workflow in detail.