Conceptify.AI MVP

Conceptify is a ChatGPT-like app that lets you build interactive concept maps from your conversations with AI - supporting meaningful learning by making it easier to visually organize ideas and recall information.

➡️ Try it out: https://conceptify-ai.vercel.app

Concept map built with Conceptify for the Linear Algebra chapter in “Deep Learning” by Goodfellow et al.

Demo

Conceptify.Demo.mp4

Features

• Create colorful, interactive concept maps from AI assistant messages that grow organically as you learn

• Enrich concept descriptions with support for LaTeX, code blocks, and image links

• Customize the look and content of your concept map with 12 color palettes and full editing control

• Test your recall with quizzes generated from your concept maps

• Compile concept maps into concise summaries that can be exported as a PDF or text file

• Free to use, no account required

• Works with both OpenAI and Anthropic models

For developers

Installation

1. Clone the repo

git clone https://github.com/julietlawton/conceptify.ai.git
cd conceptify.ai

2. Install dependencies

pnpm install

4. Set environment variables

   • Create a new file named .env
   • Set the following environment variables:
   • Required:

     OPENAI_API_KEY=<your OpenAI API key>

   • Optional:

     EMAIL_USER=<gmail email address> # Used for sending user feedback
     EMAIL_PASS=<gmail app password> #Used for sending user feedback (obtained by requesting app-specific password in gmail settings)
     UPSTASH_REDIS_REST_URL=<upstash db endpoint> #Used for demo rate limiting
     UPSTASH_REDIS_REST_TOKEN=<upstash db password> #Used for demo rate limiting

5. Run the development server

pnpm run dev

6. The app should now be up at http://localhost:3000

Future plans

Desktop App:

I envision this app becoming a Electron desktop app with persistent storage, faster performance, and the ability to run a local concept map generation model. Local generation would allow using any model for chat without API compatibility issues.

Image Uploads:

Native support for uploading and inserting images directly into cards, including drag-and-drop for screenshots. (Currently only markdown image links are supported.)

Graph Performance Improvements:

Optimizations to make the graph visualization faster, smoother, and more responsive, especially for large or complex concept maps.

Document Upload:

Support for uploading documents to chat for models that can process document inputs.

Concept Map Sharing Function:

Support for sharing concept maps, and possibly a share gallery for publicly sharing concept maps.

FAQs

Where do I get an API key?

You can get an API key from OpenAI at https://platform.openai.com or from Anthropic at https://console.anthropic.com. Both providers offer a pay-as-you go pricing.

Why aren't more model providers supported?

Currently, only OpenAI and Anthropic are supported due to inconsistent concept generation by other models. OpenAI is the recommended provider for this app, as Claude occasionally produces graphs that violate the object generation schema. I have not observed this issue with GPT-4o.

As this project evolves into a desktop app, I plan to integrate a custom model for graph generation that would run locally. This would allow you to use any model provider supported by the Vercel AI SDK for chat.

How does concept map generation work?

Conceptify uses the generateObject API from the Vercel AI SDK, which allows you to define a structured object schema and pass it along with the prompt. The model is then constrained to return a response that matches this schema.

For concept map generation, the app sends the assistant message along with the current concept map. The model is prompted to extract new concepts from the message and intelligently connect them to existing nodes in the concept map.

Why aren’t you using user accounts and server-side storage?

To keep the app completely free to use and maintain, I chose not to support user accounts or server-side storage in the MVP. Instead, all data is stored locally in your browser. There is a minimal backend for demo rate-limiting, but it's small enough to stay within free usage tiers.

In the future, I plan to turn the app into a standalone desktop app with persistent local storage — eliminating the need for accounts or cloud backend.