EduFlow

Inspiration

The main inspiration to building this was that AI chat apps have a linear history and as you send messages, the entire context and flow gets lost in the sea of messages. Also if you have a recurring prompt or do the same multi-step workflows repeatedly it quickly becomes a chore in that UX.

What it does

EduFlow transforms linear AI interactions into visual, reusable workflows. Features:

1. Visual canvas replaces linear chat history where you can see your entire pipeline at a glance
2. Node-based architecture
3. Automatic context passing. Output from one node flows into the next
4. Saved workflows. Build once, execute repeatedly with different inputs
5. Real-time execution and can refresh without the state being lost

11 AI-Powered Nodes: 📝 Essay Grader, ✅ Fact Checker, ✨ Text Improver, 📊 Text Summarizer 💡 Concept Extractor, 📅 Study Plan Generator, 🔍 Web Search 🎥 YouTube Analyzer, 🤖 Web Browser Agent via Nova Act, 📄 PDF Reader

Tech Stack

EduFlow is built using Next.js 16 and React Flow for the visual canvas, Convex for real-time state management, and Trigger.dev for background task orchestration. Vercel AI SDK was also used to connect LLMs with the text and uploaded files.

Challenges

The biggest challenge was probably execution orchestration, ensuring workflows run in the correct order and pass context between nodes reliably.

--

Discord username: ah7470

Built With

• fastapi
• next.js
• python
• react
• trigger.dev
• typescript
• vercel-ai-sdk