Inspiration
Clarity was inspired by our own experiences, and by people close to us, who live with ADHD. We noticed that many people do not struggle with ability or motivation, but with starting. When tasks pile up, everything feels urgent, and deciding what to do first becomes exhausting. Large goals often feel scary simply because they are unclear. We wanted to build a tool that removes this starting friction and replaces it with a clear, gentle first step.
What it does
Clarity helps people with ADHD start tasks and goals without feeling overwhelmed. Users can enter a list of tasks and either organize them into urgency and importance quadrants or let the AI sort them automatically. The app orders tasks in a momentum-friendly way and shows rough time estimates. Users can track progress by ticking off completed items, which creates a visible sense of accomplishment and helps lift their mood as they move forward. For big goals, Clarity breaks them into very small, concrete steps. Any step can be broken down again until it feels manageable, making progress feel achievable rather than intimidating.
How we built it
Clarity integrates the Gemini 3 Flash Preview API as the core reasoning engine that helps users with ADHD move from feeling stuck to taking action. Gemini 3 is used for task ordering, time estimation, and step-by-step planning, all of which directly address executive dysfunction and task paralysis.
For users who feel overwhelmed by many tasks, Clarity sends a list of user-entered tasks to Gemini 3 and asks the model to reorder them in a way that reduces friction. Gemini 3 reasons about momentum, starting with quick wins, alternating effort levels, grouping similar tasks, and assigning a short time estimate to each task. The model returns structured JSON using response schemas, which allows the app to reliably display ordered tasks, categories, encouragement reasons, and estimated completion times.
For large or intimidating goals, Gemini 3 is used to break a single abstract goal into very small, concrete micro-steps. The model is prompted to make the first step extremely easy and non-threatening. Users can further select any step and ask Gemini 3 to break it down again, enabling recursive decomposition until the task feels manageable.
System instructions are used to guide Gemini 3 to behave like an empathetic ADHD coach, making the AI behavior central to Clarity’s design rather than a surface-level feature.
Challenges we ran into
One major challenge was prompt design. Small changes in wording could affect tone, structure, or usefulness. Another challenge was finding the right balance between automation and user control, so the app helps without feeling pushy or rigid.
Accomplishments that we're proud of
We are proud that Clarity turns a powerful language model into a focused, supportive tool for a real problem. The recursive goal breakdown and structured task sorting work reliably and feel natural to use.
What we learned
We learned that reducing cognitive load matters more than adding features. Clear constraints, good schemas, and empathetic instructions make AI outputs far more useful in practice.
What's next for Project ADHD
Next, we plan to add personalization, progress tracking, and gentle reminders that adapt to user patterns over time, while keeping the experience calm and low pressure.
Log in or sign up for Devpost to join the conversation.