Designed for AI and Agentic workflows

Stefan Johansson · 23 May 2026 · 2 min read

ADL runs locally and stores metadata as plain, readable files. That makes it a natural fit for AI-assisted development: your editor, chatbot, or agent can inspect the same project files your team already works with.

*This image was generated using AI, in VS Code, using the same type of AI-assisted workflow described in this post.

Why this matters

Because ADL projects are file-based, AI tools can reason over metadata definitions, templates, generated outputs, and documentation without needing a proprietary integration or remote service.

What you can do

• Enrich your metadata definitions.
• Suggest classifications, descriptions, and relationships.
• Create or refine code templates from business requirements.
• Review generated outputs for consistency.
• Ask natural-language questions about your model.

and so much more.

Still under your control

AI can accelerate the workflow, but ADL keeps the source of truth local and reviewable. Changes can be inspected in your editor, compared in Git, and validated before they become part of your project.

This means you can use your favorite AI tools to interact with your metadata, generate templates, and even build agents that automate parts of your workflow.

Your favorite AI tools

Open the metadata repository in VS Code, or your favorite AI-enabled editor. Point Copilot, Codex, Claude, or Gemini at the files. This works straight out of the box, no special integration needed.

The metadata files are designed to be human-readable and AI-friendly, so you can easily understand and manipulate them with your tools of choice. This gives you the freedom to integrate ADL into your existing workflows and leverage the power of AI to enhance your productivity.

ADL also generates AI-friendly text files, which can feed into the AI workflows, allowing you to validate, analyze, refine, and iterate on outputs as well as inputs.

Open formats

Since ADL is based on open formats, local files, well-known conventions and Handlebars, an established templating language AI easily understands, it is easy for your agentic workflows to interact with and understand your project.

Human in the loop

The ADL app loads and refreshes metadata from these same metadata files, so any changes made by AI tools can directly be seen in the app. All changes can be reviewed and approved both in the app, in your editor, and in Git, giving you full control over the evolution of your project. This makes it easy to integrate AI into your workflow while maintaining oversight and quality control.