Prompt tooling
Shipped product with a Chrome extension and subscription tiers
VortisAI
Better prompts, wherever you type them. Enhancement, query fixes, a templates library, and a Chrome extension that follows you into ChatGPT, Claude, and Gemini.
Problem
The gap between a vague prompt and a useful answer is mostly the prompt. People know what they want from ChatGPT, Claude, or Gemini and still get mediocre output because the request was underspecified. Rewriting it by hand every time does not scale.
Approach
VortisAI sits between the person and the model. Standard enhancement rewrites a prompt for clarity and structure, fix query repairs a prompt that is not working, and selective fix touches only the part you point at. A templates library covers the recurring cases.
A Chrome extension puts all of it inside the AI tools people already use, the Next.js app and Node.js backend run the enhancement, and Firebase handles accounts and subscription tiers.
Architecture
1. Capture the prompt
From the web app or straight inside ChatGPT, Claude, and Gemini via the extension.
2. Enhance
Standard enhancement, fix query, or selective fix, run by the engine through an LLM.
3. Reuse
The templates library covers the prompts people write over and over.
4. Accounts and tiers
Firebase handles auth, data, and subscription tiers.
Stack
- Next.js
- Node.js
- Firebase
- Chrome extension
- LLM
Outcome
A shipped product with the web app, the extension, the templates library, and paid tiers, working across the major AI models.
What I'd do differently
The extension mattered more than the web app. Enhancement people have to copy and paste into another tab gets used once; enhancement that happens where they are already typing gets used daily.