Semanticus
Version 1.0.1 · free core · cross-platform

Build, test and ship semantic models. All in one VS Code workbench.

Semanticus works with Power BI and Fabric semantic models. Edit a model, check important numbers, understand what a change will affect, and review the exact difference before you publish. Your AI Assistant can work in the same session, with every change visible, attributed and undoable.

Install for VS Code, free See Pro: US$79/yr or US$10/mo
Five platform packages for Windows, macOS and Linux  ·  no telemetry
Semanticus Studio showing a semantic model diagram in VS Code

One place for the whole model journey

Open a PBIP, TMDL or BIM model, a running Power BI Desktop model, or a remote XMLA model. Work through the same Studio from first edit to final review.

Build and edit

Edit tables, measures, columns, relationships, calculation groups, security, DAX and M. Use modern calendars, model properties, a guided model spec and a full undo history.

Understand

Explore the diagram, preview data, inspect storage, search the model and trace lineage. See what depends on an object before you rename or remove it.

Improve

Run AI Readiness and Best Practice scans. Fix findings one at a time, write your own rules and keep accepted findings visible with a recorded reason.

Prove

Reconcile SQL and DAX, run repeatable model tests, interview the model and save evidence with the model. Prove a DAX rewrite returns the same values before applying it.

Review and ship

Compare models, review selected changes, check the release gate and preview every cloud write. Published pushes create a restore point before they change the target.

Work locally and publish deliberately

Keep separate identities for the model you edit, the model used by tests and the final publish destination. Prepare a safe local working copy when the source is remote.

Your AI Assistant works in the same session

You use the Studio. Your AI Assistant uses MCP. Both work on the same model and share the same undo history.

Watch every change as it happens

Connect an MCP-compatible AI Assistant, then ask it to document measures, investigate unused columns, improve a DAX expression or work through readiness findings. Its changes appear in the Studio and model tree immediately.

  • 302 typed operations across the model journey
  • One shared timeline with every change attributed to You or AI Assistant
  • Human-only confirmations remain human-only
  • The engine runs no inference and holds no API keys. Your assistant, your account, your data
Edit History showing changes made by the user and AI Assistant
The Tests workspace showing repeatable model checks

Test the model, not just the code

The Prove workspace brings Tests and Evidence together. Reconcile trusted SQL totals with DAX, check important measures across contexts, replay model interviews and export a report that records what passed.

  • Save reusable tests beside the model
  • Trace a failed dependent measure back to its root failure
  • Seal test reports and workflow results as evidence
  • Keep the complete record under source control

Use a workflow when the change is difficult

Workflows turn a job into a clear sequence of questions, edits and checks. The built-in hard-measure workflow uses one candidate, an independent raw-row answer, a context grid and a hard equivalence check before optional performance work.

  • 14 focused built-in workflows
  • Profiles select a complete workflow policy in one action
  • Required steps and overrides are recorded
  • Completed runs can export sealed evidence
An enforced workflow with questions and verified gates

Get Semanticus

Free includes the workbench, individual edits, scans, tests and previews. Pro adds reviewed bulk changes, enforced workflows, Verified Mode, continuous evidence and Pro-gated cloud writes. Compare Free and Pro.

Install free from the VS Code Marketplace View on GitHub
Does it work with my setup?

Version 1.0.1 has packages for Windows 11 x64 and ARM64, Ubuntu 24.04 x64, macOS Intel and macOS Apple Silicon. Offline model files and remote XMLA work across those packages. Power BI Desktop discovery, local XMLA and local M preview require Windows 11 x64. Read the support details.

Is my model data sent anywhere?

No. The engine runs entirely on your machine, holds no API keys, performs no inference and sends no telemetry. AI work happens through your own AI assistant on your account, connected over MCP.

What exactly is Pro?

Pro is the enforcement and evidence layer. It adds reviewed bulk changes, Verified Mode, enforced workflow runs, advanced evidence, continuous health checks and Pro-gated Data Agent writes. See the full comparison.

With gratitude to open source

Semanticus is built with open-source projects that make this work possible.