What is Magic Potion?
Magic Potion is a visual workspace for building and testing AI prompts. Instead of editing text files, users drag system messages, user inputs, and assistant responses onto a canvas, organize them into projects, and run the prompt with results displayed inline. The page positions it for use across desktop, tablet, and phone. Live page returned minimal content during research, so claims track the page description.
Why Magic Potion works
Maintaining a serious prompt library quickly devolves into scattered text files and Notion pages where structure gets lost. A visual canvas with typed blocks for each message turns the prompt into something that can be inspected and edited at a glance, which speeds up iteration for anyone tuning prompts as a core part of their work.
Magic Potion features
- Block-based prompt canvas. Build prompts by dragging system, user, and assistant blocks instead of editing raw text.
- Inline test runs. Run prompts and see results in the same view without copying between tabs.
- Project organization. Group prompts by project so libraries stay structured as they grow.
- Cross-device access. Works on desktop, tablet, and phone for working on prompts away from a workstation.
Who Magic Potion is for
- AI engineers maintaining a library of prompts across products who need version-able, inspectable structure.
- Marketing operators iterating on content generation prompts for blogs, ads, and emails.
- AI researchers running quick experiments across model providers and tracking variants.
- Solo founders building AI features who want their prompt logic outside their codebase while still under control.
Similar micro SaaS ideas you can build
- Prompt eval suite for solo devs. A lightweight tool that runs a prompt against a saved test set and tracks accuracy across edits, sold per developer seat.
- Prompt library for agencies. A shared workspace for marketing agencies that catalogs client-specific prompts with permissions, billed per agency.
- Prompt diff and review. A tool that diffs prompt versions and supports approval workflows before production deploy, sold per AI team seat.
