YasnaMCPThe bridge for local business software
ENفارسیالعربية
Dashboard login
← Back to YasnaMCP
YasnaMCP Knowledge Hub

Design and Publish Power BI Reports with AI and YasnaMCP

In the new YasnaMCP Windows Agent, Power BI is more than a separate reporting tool. A connected AI client can inspect databases, understand their structure, design a Power BI report, view actual Power BI Desktop output, refine it and publish it to Power BI Report Server.

Power BI + AI + YasnaMCP

Design, view and publish Power BI dashboards with AI through YasnaMCP

In the new YasnaMCP Windows Agent, Power BI is more than a separate reporting tool. A connected AI client can inspect databases, understand their structure, design a Power BI report, view actual Power BI Desktop output, refine it and publish it to Power BI Report Server.

Current version: This capability has been available since YasnaMCP Windows Agent 1.4.3 and was designed and tested for Power BI Desktop for Report Server and Power BI Report Server.
AI Client → MCP → YasnaMCP Gateway → Windows Agent → Power BI Desktop → Power BI Report Server

What actually happens?

Instead of manually building data models, measures, relationships, visuals, themes and publishing in Power BI Desktop, a user can describe the requirement in natural language. YasnaMCP exposes Power BI tools to the model, and Windows Agent performs the actual operations on the user’s machine.

  • Analyze the database and identify fact tables, dimensions, keys, dates and financial values.
  • Run read-only queries to inspect data before dashboard design.
  • Create reports, tables, columns, relationships and measures.
  • Generate a Jalali calendar and Persian sales measures.
  • Design pages and visuals on a standard 12 × 12 grid with RTL support.
  • Apply ready-made themes for sales, finance, inventory and management dashboards.
  • Open reports in Power BI Desktop, load data and capture page images automatically.
  • Return actual page images to AI so it can review and refine the result.
  • Save the PBIX file and publish directly to Power BI Report Server.

AI does not just generate a file; it reviews the result

The design cycle does not stop at file generation. Power BI Desktop actually runs, loads the data, renders the report and captures each page. Images return to the model through MCP, allowing it to decide what needs changing based on what Power BI actually rendered.

Agentic cycle: Analyze → Author → Render → See → Improve → Render Again → Publish

Power BI without patching Desktop

This architecture does not patch Power BI Desktop. YasnaMCP uses documented Power BI formats such as PBIP, TMDL, PBIR, PBIT and Custom Visual Packages. Desktop handles tasks Power BI itself must perform: loading data and rendering reports.

  • The primary YasnaMCP project is stored as yasna-report.json a project document.
  • The same document produces PBIP output containing TMDL and PBIR.
  • A PBIT file is also produced for Power BI Desktop for Report Server.
  • After rendering, the final PBIX is saved and can be published.

Designed for Iranian workflows: Jalali dates, RTL, rials and tomans

YasnaMCP includes a design system in the Agent for Persian reporting scenarios, alongside Power BI automation.

  • RTL layout: Because Power BI has no page-wide RTL switch, visuals are positioned using a mirrored grid.
  • Jalali calendar: A Persian date table is generated in DAX, with leap-year logic based on PersianCalendar.
  • Jalali Date Picker: A dedicated custom visual supports selecting a Jalali date, date range or YYYY/MM/DD input.
  • Rials and tomans: Stored and displayed units are separate, and currency measures can apply suitable conversion and labels.
  • Persian fonts: IRANSans FaNum is installed with the Agent to display Persian digits in native visuals.

Ready-made templates and themes

AI does not need to design each dashboard without a framework. YasnaMCP provides themes and templates for structured, repeatable output.

  • yasna-light For general light dashboards.
  • yasna-dark For dark management dashboards.
  • yasna-finance For financial reports.
  • yasna-warehouse For inventory reports.
  • Templates: sales-executive, sales-daily, finance-overview and inventory-overview.

30 Power BI tools available to AI

The gateway exposes Power BI tools through three separate security scopes, keeping reading, authoring and publishing distinct.

  • powerbi.read: View status, reports, templates, images, database analysis and read-only queries.
  • powerbi.author: Create and edit reports, tables, measures, relationships, pages and visuals, and run rendering.
  • powerbi.publish: Publish to Power BI Report Server.
Access control: The connector’s database allow-list applies to the database used by the report, preventing access to databases outside the configured scope.

An example dashboard workflow

powerbi_status → analyze_database → query_data → create_report → add_table → relationship → measures → Persian Calendar → template → render → inspect image → improve → publish

The model first checks machine capabilities, analyzes the database, inspects data, creates a report and reviews rendered output. The same report can then be published to Report Server.

Connecting to Power BI Report Server

The Agent can discover Report Server on the machine, create the destination folder, upload a PBIX and store data-source credentials on the server. A dedicated Power BI settings tab covers database, Desktop, Report Server, theme, RTL, currency and Jalali range.

What still requires the user once?

When Desktop first connects to a database server using SQL authentication, Desktop requests the password. YasnaMCP deliberately does not fill that password dialog for the user. The Agent returns status waiting_for_credentials and an image of the prompt. Once the user enters the password, automation continues. Desktop remembers the credential for later use.

What was tested in the reported version?

  • Power BI Desktop for Report Server, September 2026 edition.
  • Power BI Report Server on the same machine.
  • Two actual SQL Server databases.
  • Analysis, authoring, validation, rendering, PBIX saving and publishing.
  • Persian layout, Jalali axes, Persian digits, Jalali date picker and actual filtering on Report Server.
  • Storing data-source credentials on Report Server during publication.

Items not yet reported as verified

  • PBIP output in the cloud edition of Power BI Desktop.
  • Complete rendering with the Agent running only as a Windows Service without a user session.
  • Viewing a published report on a machine without the Agent’s Persian font installed.

What does this mean for business?

YasnaMCP makes Power BI usable by an AI agent. A manager or finance user can describe a requirement in natural language, and AI can use controlled data access, a defined design system and an actual render/review cycle to build and publish a dashboard on the organization’s on-premises infrastructure.

A final scenario: ‘Build a Persian management dashboard from this year’s sales with a Jalali calendar, daily sales, branch comparisons, amounts in tomans and a date-range filter, then publish it to Report Server.’