Vibe Coding a Desktop Tool with AI

Video by: Reid Havens

This project was built almost entirely using Claude Code through pure vibe coding. No detailed specs.

No formal architecture. Just me giving the LLM ideas, screenshots, and plain-English requests until a real desktop tool existed. In this video, I walk through the Desktop Widget Wall I built and focus on the art of the possible, not the implementation details. This is a non-technical walkthrough meant to show how far you can go when you treat an LLM like a creative collaborator instead of a code generator.

You’ll see how:

• An idea turns into a working desktop tool

• Screenshots and rough concepts become real UI

• AI can iterate on structure, layout, and behavior - “Vibe coding” can replace traditional dev workflows for small tools

• You can prototype software without writing much code yourself This isn’t a tutorial and it’s not a “how to prompt” video.

It’s a real example of what happens when you lean into AI-assisted building and let the tool do the heavy lifting. If you’re curious about rapid prototyping, personal tooling, or what modern LLMs can actually build, this should give you plenty of ideas.

RELATED CONTENT 🔗

Widget Wall Repo

Power BI to Plotly: Build AI Dashboard Apps - with Adam Schroeder

LIVESTREAM DATE/TIME 📅

March 6th - 9:30 AM (Pacific Time)

DESCRIPTION 📄

You’ve already invested in a Power BI semantic model. Cool. Now let’s actually use it outside the walls of a closed BI environment. In this session, Adam Schroeder (Plotly) shows how Plotly Studio can connect directly to your existing Power BI semantic model so you can build polished, production-ready data apps, powered by Python and accelerated with AI, without rebuilding or migrating your data.

You’ll learn how to:

• Build fully customized, production-ready data apps using Python (beyond drag-and-drop limits)

• Distribute apps broadly without getting crushed by per-user viewer licensing costs

• Own your branding, UX, and deployment strategy end-to-end

• Take a practical step toward operationalizing trustworthy AI on top of governed data

If you’re hitting the ceiling on Power BI customization, navigating licensing-based scaling challenges, or looking for more control over how insights (and AI) reach users, this is your playbook.

GUEST BIO 👤

Adam Schroeder is a Senior Developer Advocate at Plotly. He has helped grow the community through 1:1 relationships, contributing to 6 million Dash downloads per month. He’s taught Plotly Dash for over five years on YouTube as @CharmingData (over 60,000 views/month) and co-authored “The Book of Dash.”

RELATED CONTENT 🔗

Plotly Studio

AI-Powered Power BI: Building Custom Themes & Visuals with GitHub Copilot - with Zoe Douglas

LIVESTREAM DATE/TIME 📅

February 20th, 2026 - 9:30 AM (Pacific Time)

DESCRIPTION 📄

GitHub Copilot can turn Power BI customization from a tedious, manual grind into a fast, interactive workflow. In this session, you’ll see how to generate polished report themes by extracting color palettes from images, create named style presets that show up directly in Power BI’s formatting pane, and build custom visuals from scratch with Copilot as your pair programmer. From JSON theme schemas to visual implementation details, you’ll leave with practical techniques to speed up development, improve consistency, and ship better-looking reports with less trial-and-error.

GUEST BIO (Zoe Douglas)👤

I'm a product manager for Power BI working on core visuals with focus on the default and theming experience with visuals in reports. I'm also good with a semantic model and worked on features such as DAX query view, Copilot to write DAX queries, Copilot to write measure descriptions, Optimize ribbon, and more! Before coming to Power BI product group I was a Power BI consultant using Power BI everyday to bring insights to action!

RELATED CONTENT 🔗

LinkedIn
Zoe's Blog
Zoe's GitHub

Power BI Report Cleanup Just Leveled Up (Duplicate Images, Orphans, Saved Scripts)

Video by: Reid Havens

Power BI reports quietly accumulate junk. Duplicate images. Orphaned files. Saved DAX and TMDL scripts that were never meant to ship.

In this video, I walk through the Report Cleanup enhancements in AE Power BI Multi-Tool Suite v2.0, focusing on how the tool helps you clean development artifacts safely and measurably before publishing.

You’ll see how the tool:

• Detects duplicate images using MD5 hashing and consolidates references

• Finds and removes unused and orphaned images across PBIP files

• Cleans up saved DAX queries and TMDL scripts left behind by Desktop tooling

• Reports exact size savings so you know what you actually gained

These are artifacts created during real development workflows. They’re invisible in Desktop, but they absolutely matter for repo hygiene, file size, and long-term maintainability.

This is about shipping cleaner, leaner Power BI reports without guessing or manually editing JSON files like a raccoon in a data dumpster.

RELATED CONTENT 🔗

AE Power BI Multi-Tool
AE Website