Lingollo

• 

  Homepage

About the project

Lingollo is an AI-powered web application that helps French learners improve their writing skills through real-time, structured feedback. It combines language pedagogy with AI to deliver an experience similar to Duolingo + Grammarly, focused on the A1 CEFR level.

Inspiration

While learning French, I found a clear gap—most platforms focus on reading, listening, or grammar drills, but none provided structured writing practice with instant, level-appropriate feedback. That’s when I decided to build Lingollo: a tool that guides learners through daily writing themes, giving them actionable corrections and confidence as they progress.

How we built it

Platform: Built using Bolt.fun, a no-code AI builder, for fast prototyping and API integration.

Backend: Connected to Supabase for storing writing data, user sessions, and tracking progress.

AI Layer: Integrated OpenAI GPT-4 API with custom system prompts tailored to A1-level French writing.

Curriculum Design: Based on 20 CEFR A1 themes, each with 5+ targeted writing prompts. Learners get grammar, vocabulary, and spelling feedback with guided corrections.

Frontend Features:

Writing input box and instant feedback loop

Progress tracking (theme completion, prompt history)

User streak model (planned in upcoming phase)

Challenges we ran into

Prompt Engineering: GPT was too advanced for A1 users initially. I had to fine-tune system prompts, control vocabulary, and introduce a strict grammar+vocab JSON.

Token Efficiency: Early prompt designs consumed too many API tokens. I restructured them to minimize cost while preserving learning quality.

Accomplishments that we're proud of

🎯 Designed a CEFR-aligned A1 writing curriculum with real-life themes and guided prompts.

⚙️ Built the entire app using Bolt and connected it with Supabase to store writing input, user history, and theme progress securely.

🧠 Engineered a custom GPT system message that delivers A1-level grammar, spelling, and vocabulary feedback in real time.

📊 Created an internal grammar and vocabulary JSON dataset to guide consistent AI feedback across all themes.

🧪 Conducted early user testing and validated writing engagement patterns, leading to a pivot from passive lessons to interactive AI writing practice.

🚀 Launched a working MVP with writing streak tracking, prompt sequencing, and progress control—all without requiring users to have a ChatGPT subscription

What we learned

How to scope and design a curriculum-driven AI app focused on one skill (writing).

The power of micro-feedback for language learning.

Hands-on experience with Bolt + Supabase integration, including data schema design and API chaining.

The importance of prompt control and UX flow when using AI in education.

What's next for Lingollo

Launching user dashboards with writing analytics, revision prompts, and performance badges.

Teacher dashboard for assigning writing tasks and reviewing AI-generated feedback.

Optional gamified elements (streaks, goals, revision reminders)

Built With

• react