FocusFlow AI

Inspiration

Starting an overwhelming academic task can be harder than completing it. We wanted to build an AI tool that focuses on the moment before productivity begins: helping a student take the first step.

FocusFlow was designed around a simple idea: instead of asking AI to complete a student's work, use AI to reduce the barrier to starting.

What it does

A student describes something they feel overwhelmed by. FocusFlow uses Gemini to identify one concrete action that can be completed in just a few minutes.

If the student clicks "I'm stuck," FocusFlow adapts and generates an even smaller action.

This creates an interaction loop:

Overwhelming task → Small first step → Student feedback → Even smaller step

The goal is to make AI feel supportive rather than overwhelming.

How we built it

FocusFlow uses a JavaScript frontend connected to a Node.js and Express backend. The backend securely communicates with the Google Gemini API and sends the student's task to the model with instructions to generate a single, achievable starting action.

We also designed the system so the Gemini API key remains on the backend rather than being exposed in the frontend.

What we learned

We learned how to build an AI-powered application from the ground up, including API integration, backend development, environment-variable security, frontend/backend communication, and adaptive prompting.

One of the biggest lessons was that an AI application does not need to solve an entire problem to be useful. Sometimes the most valuable output is simply helping someone begin.

Challenges

A major challenge was designing the AI response so that it did not overwhelm the user with another long list of instructions. We addressed this through focused prompting that asks Gemini to provide only one concrete first action.

We also built an adaptive "I'm stuck" mode so the system can respond to the user's feedback instead of repeatedly giving the same type of answer.

What's next

Future versions could learn from user feedback, provide optional focus timers, support additional accessibility preferences, and evaluate whether smaller AI-generated steps actually improve task initiation and completion.

Built With

• ai
• css
• education
• express.js
• gemini
• html
• javascript
• ml
• node.js