Vibe Coding a Power BI Embedded Portal - From Zero to Hero in 60 Minutes
📎 Slide deck: Vibe Coding a Power BI Embedded Portal - Peer Gronnerup.pdf
Speaker: Peer Grønnerup — Technical Architect & Community Advocate @ Tabular Editor
15+ years BI/Data Platform, software developer roots, "strategically lazy" (automates everything)
🔗 LinkedIn | Blog | GitHub
Key Takeaways
- Power BI Embedded lets you fully white-label and control the report experience for internal or external users
- Two embedding models: App owns data (external, any F/P/A SKU, no user licenses) vs User owns data (internal, Entra ID, needs Pro/PPU or F64+)
- Vibe coding = describe what you want in natural language, AI builds it — no heavy upfront design
- Domain knowledge is the multiplier — knowing your tech stack makes AI output dramatically better
- Vibe coding is for prototyping and iteration, NOT a replacement for architecture best practices
- Start simple: conceptualize first, iterate fast, expect to debug
Notes
What Is Power BI Embedded?
- Embed Power BI reports into your own applications
- Supports internal and external audiences
- Full control and white-labeling of the experience
Embedding Options
| Feature | External Users | Internal Users |
|---|---|---|
| Authentication | App managed | Microsoft Entra ID |
| Licensing | F, P, or A SKU* | Pro, PPU, P, or F64+ |
| Primary use case | External-facing apps | Internal business apps |
| Ownership | App owns data | User owns data |
| Implementation | Custom coding | Secure embed (iframe) / custom coding |
*No user licenses required for external (app owns data)
Why Embed?
- Better user adoption, seamless UX (branding + layout)
- Control access & navigation
- Can be low-code or no-code
- Cost savings — use internally or externally (intranet, customer portal)
Cons of Embedded Analytics
- Requires capacity = complex pricing and capacity planning
- More engineering effort
- Operational complexity: multi-tenant design, lifecycle management, authentication, hosting
- Feature parity nuances vs. native Power BI
Vibe Coding
What it is:
- AI-powered development practice
- No/low technical skills required — describe ideas in natural language, AI builds it
- Shift from writing code to creating products through intent
- Let energy, ideas, and intuition lead
Pure vibe coding vs. responsible AI-assisted development — spectrum, not binary
Pros:
- Speed, creativity, easy iteration, accessibility, fast prototyping
Dangers of pure vibe coding:
- Messy code, no long-term structure, hard to debug
- Loss of understanding, security gaps
Session Goal
Build a Power BI Embedded Portal live:
- Targeted at externals (App owns data)
- Login/authentication
- Information pages
- Dynamic RLS support
- Additional UX: fullscreen, page navigation
Target: fully production-ready, deployed and running in Azure
Prerequisites / Toolbox
Tools used:
- VS Code + GitHub Copilot
- Context7 MCP — https://github.com/upstash/context7
- Azure + Azure App Service
Power BI prereqs:
- Workspaces with semantic model (no roles), reports for demo
- Service Principal with client credentials
- GitHub Copilot instructions file: https://tinyurl.com/yw5f85ht
Final Tips & Takeaways
- Conceptualize first — start with a clear idea of what to build
- Domain knowledge is power — knowing relevant frameworks improves AI output significantly
- Iterate — expect to debug and refine; AI is not a magic wand
- Use with caution — vibe coding for testing ideas and prototyping only
- Not a replacement for architects and best practices
Recommended Vibe Coding Tools
| Tool | Best For |
|---|---|
| VS Code + GitHub Copilot | Developers wanting AI in their existing IDE |
| Cursor / Windsurf | Pro devs wanting an AI-first IDE |
| Lovable | Non-coders and low-coders (most user-friendly) |
| Claude Code | Experienced devs, complex multi-file projects |
Peer's starter repo: github.com/gronnerup