Power BI as Code: PBIP & the New VS Code Integration

Power BI is quietly becoming code — and the August 2026 update is a real milestone on that path. Alongside the visual and modeling features, Microsoft added VS Code integration and instant reloads for Power BI Project (PBIP) development. If you build serious models, this changes how you can work. Here is what it means, and why it matters.

First, what is a Power BI Project (PBIP)?

Traditionally, a Power BI file is a single binary .pbix — great for sharing, terrible for version control, code review or automation. The Power BI Project (PBIP) format changes that. Instead of one opaque file, PBIP saves your work as a folder of plain-text files:

• The semantic model as TMDL (Tabular Model Definition Language) — your tables, columns, measures and relationships, defined in readable text.
• The report as PBIR — the pages, visuals and formatting, stored as structured JSON.

Because everything is text, it can be diffed, reviewed, branched and tracked in Git — exactly like software. PBIP is the foundation everything below is built on.

What’s new in August 2026

Two practical improvements landed this month:

• Instant reloads. Power BI Desktop now detects when PBIP project files have been changed outside Desktop — by an editor, a script or an AI agent — and prompts you to apply those changes with a single click. Previously this meant closing and reopening the file.
• A built-in VS Code entry point. Power BI Desktop can now open the project directly in Visual Studio Code, so you can move between the visual designer and code editing without hunting for the folder yourself.

Small on the surface, but together they remove the friction that used to make code-based Power BI development feel like swimming upstream.

Why edit Power BI in VS Code?

Once your model is TMDL and your report is PBIR, a proper code editor gives you things Power BI Desktop simply cannot:

• Fast find-and-replace across the whole model — rename a column or fix a pattern in every measure at once, instead of clicking through the fields pane.
• TMDL syntax highlighting and editing via extensions — author measures and calculation groups as text, with structure you can actually read.
• Bulk edits and scripting — generate or refactor dozens of measures programmatically rather than one dialog at a time.
• Git built in — stage, commit, branch and review right where you are editing.

The real unlock: Git-based development

This is where PBIP earns its keep. Because the model and report are text, a Power BI project can live in a Git repository like any codebase — which means:

• Branching for new features and fixes, without touching production.
• Pull requests and code review — a teammate can actually see, line by line, which measures and relationships changed before it ships.
• A full history — who changed what, when and why, with the ability to roll back.
• CI/CD — automated validation and deployment with tools like the Tabular Editor CLI and GitHub Actions, and, in Microsoft Fabric, workspace-level Git integration and deployment pipelines.

For teams used to emailing Report_final_v3_FINAL.pbix around, this is a different world.

And the bigger picture: AI-assisted development

Text-based projects plus a code editor are exactly what AI coding agents need. With PBIP open in VS Code, an agent like Claude — connected to your model through a Model Context Protocol (MCP) server — can batch-create measures, audit relationships, and enforce your naming conventions against a real model, with every change tracked in Git. Power BI Copilot handles first-draft DAX and narratives inside the report; the agentic workflow handles structured, reviewable model changes outside it. This is the exact toolchain we teach in our Power BI in the Age of AI course — MCP, PBIP, VS Code, GitHub and Copilot, end to end.

How to get started

1. In Power BI Desktop, save your work as a Power BI Project (File > Save as, choosing the .pbip format; enable it under Options and settings > Preview features if you don’t see it yet).
2. Use the new entry point to open the project in VS Code, and add the TMDL extension for a better editing experience.
3. Run git init in the project folder (or push it to GitHub) and commit your first version.
4. Make a change in VS Code, save, and let Desktop’s instant reload pull it back in — you now have a two-way, code-first workflow.

Who this is for

If you are a solo analyst building a couple of reports, .pbix is still fine. But if you are a BI developer or a team shipping models that matter — where review, history, standards and automation are the difference between reliable and chaotic — PBIP with VS Code and Git is how modern Power BI development is done.

This was one of several developer updates this month — see the full Power BI August 2026 update for the rest.

Source: Microsoft — What’s new in Power BI, August 2026 and the Power BI Projects overview.