ThinkDeep

Landing Page
Choose your coach
Chose thinking mode
Go through your thinking process (Keyboard/Mic input)
Further going into thought process
Running it through ThinkDeep tailored LLM (using OpenAI API)
User feedback on thinking
User feedback on thinking
Dashboard that shows statistics
Graph can me modified to show different thinking stats over time
Achievements

Inspiration

We built ThinkDeep because we noticed that people consume a lot of information every day but rarely stop to question it. Social media, news, and online debates move very fast. Most people react emotionally instead of thinking critically. We wanted to create something that does not just give answers like a normal AI tool, but actually trains people to think better. The idea was to build a digital thinking coach that guides users step by step through reasoning and reflection.

What it does

ThinkDeep is an AI powered critical thinking app. Users can enter a claim, a decision they are trying to make, or a situation they want to analyze. The app then takes them through structured thinking. It breaks down assumptions, shows different perspectives, highlights possible bias, and asks reflection questions. Instead of giving a final answer, it push users to think deeper. In the app, users earn XP when they reflect, identify assumptions, and engage with counterarguments. There is a level system, streak tracking, and a depth score that shows how strong their reasoning is. The goal is to train the mind like a muscle.

How we built it

We built ThinkDeep using Next.js and TypeScript for the frontend. For styling we used TailwindCSS to keep the design clean and modern. We used Supabase as our database to store users, analyses, reflections, XP, and levels. For the AI functionality, we integrated the OpenAI API. We designed structured prompts so the AI would respond in a clear format. Instead of asking the AI to just answer questions, we instructed it to act as a critical thinking coach. It returns structured analysis including assumptions, counterarguments, bias detection, and reflection questions. We also implemented a scoring system where the AI evaluates the depth of the user’s reasoning and gives feedback.

Challenges we ran into

One of the biggest challenges was making sure the AI does not simply give direct answers. We had to carefully design the prompts so that the AI encourages thinking instead of replacing it. Another challenge was building a fair depth scoring system. Measuring reasoning quality is not simple. We had to experiment with different formats and scoring logic. We wanted it to feel engaging and motivating without turning deep thinking into something childish.

Accomplishments that we're proud of

We are proud that ThinkDeep does more than summarize or fact check. It actually takes the user through structured reasoning. The XP and level system adds motivation and makes users want to improve. The guided reflection flow is something we believe makes the app unique compared to many other AI tools. We are also proud of the clean UI and smooth user experience. The app feels simple but has so many cool features.

What we learned

We learned that building with AI is not just about calling an API. Prompt design is extremely important. Small wording changes can completely change how the AI responds. We also learned that encouraging critical thinking requires careful UX design.If you guide them step by step, users participate more deeply. Finally, we learned how to integrate AI with a real database system and create a full stack application under 24 hours.

What's next for ThinkDeep

In the future, we want to improve the scoring system and make it more personalized. We also want to add daily challenges and maybe a classroom mode where teachers can use ThinkDeep with students. Another idea is to create a browser extension that helps users analyze articles directly while reading them. Our long term goal is to make ThinkDeep a platform that helps people build strong reasoning skills.