How RASON Desktop Works

RASON Desktop connects models created in RASON with Microsoft Power BI, allowing them to be executed directly within a report. Instead of running models in a separate environment, users can interact with inputs and view results within the same Power BI interface.

The overall workflow can be summarized as follows:

  1. Build and test a model: Create and validate a model in VS Code using the RASON modeling language.
  2. Configure data input and output within the model: Define how the model imports data from Power BI and/or Excel and exports results for reporting and visualization.
  3. Export the model to Power BI or Excel: Use Frontline’s Visual Studio Code extension to export the model into a Power BI report or an Excel workbook.
  4. Run the model within the report: Trigger the model to solve directly in Power BI or Excel.
  5. Display and analyze results: Results are returned to Power BI or Excel and visualized using tables, charts, and other report elements.

This workflow allows users to move seamlessly from model development to interactive analysis without leaving the Power BI or Excel environments.

Power BI and Excel Integration

The extension connects directly to running instances of Power BI Desktop and Excel Desktop, letting you read inputs and write results without leaving Visual Studio Code.

  • Auto-discovery — Open Power BI Desktop reports and Excel Desktop workbooks are detected automatically and shown in the RASON sidebar.
  • Read from tables — Bind RASON data sources to:
    • Power BI tables, columns, or measures, listed for you in the RASON sidebar.
    • Excel cell ranges by dragging a worksheet or range into your model.
  • Pick tables interactively —
    • Drag Power BI data into your model to auto-generate a Power BI data source, with the DAX query written for you.
    • Select a cell range in Excel and have its address written for you into your model.
  • Write results to Power BI —
    • After solving a model, export any result data directly to a Power BI table with a single click.
    • Export any result data from the Solve Console directly into an Excel cell range.
  • Live updates —
    • Charts and visuals in Power BI refresh automatically, preserving your own columns, labels, and categories.
    • Excel charts linked to your data refresh automatically, preserving your chart annotations.
  • Embed models in reports — Save the full RASON model inside the Power BI report or Excel workbook so it travels with the file and can be reused later.

Solver Modes

The extension includes a collection of powerful solvers capable of handling the full range of RASON models, including mathematical optimization, simulation and risk analysis, data science and machine learning, business rules, and multi-stage decision flows where one model’s solution feeds into the next.

Licenses for desktop and cloud solvers are available from Frontline Systems. You can also solve in the cloud directly from Visual Studio Code on your desktop, using Frontline’s hosted RASON service and its REST API.

Three solver modes are available from the status bar or the Toggle Solver Mode command:

  • Local — Use the bundled solvers. This is the default mode.
  • Cloud Synchronous — Submit the model to rason.net and wait for the result.
  • Cloud Asynchronous — Submit long-running models and poll for completion.

Cloud solving requires an authorization token, which can be set using RASON: Set Auth Token. See the Installation chapter for more information.

MCP Server Makes You a RASON Expert!

The included MCP Server gives you a Copilot "RASON Expert". RASON Desktop includes the RASON MCP Server, giving GitHub Copilot Chat full access to RASON's modeling, analysis, and solving capabilities. It's more than a REST API wrapper — the server includes a searchable library of nearly 200 example models, templates that guide model generation, structural analysis, and solve result interpretation.

Open Copilot Chat and start asking questions!  Example search, templates, model analysis, and RASON info work immediately. Solving and model management require the same authorization token as cloud solving.