Overview

‍

‍

[[symbol=divider]]

AI Hypothesis

If AI can have a deep enough understanding of a user’s lifestyle, goals, habits, and schedule such that it can suggest an optimal schedule for the user in the form of small, easy-to-complete activities with an over 80% accuracy, and the user is in a position to adhere to the suggested schedule, which in turn reduces their level of stress and overwhelm, freeing the time and energy normally spent planning their day to now allow them to accomplish their goals and be their most creative selves, our company can capture a fee from subscriptions.

‍

‍

[[symbol=divider]]

‍

Customer Research

After interviewing professionals with and without ADHD, using a carefully-crafted interview script  we came up with valuable insights:

Paint Points:

• Feeling overwhelmed with large, ambiguous tasks and need help to stay structured.
• Difficulty with multitasking and context switching, causing delays and decreased productivity.
• Struggling with prioritization and accountability, often forgetting to schedule tasks or feeling ashamed after missing a commitment.
• Difficulty maintaining focus and avoiding distractions, especially from social media and phone notifications.

‍

Unmet Needs

User interviews, combined with a competitive analysis,  can be summed up into clear unmet needs in the productivity market.

• ‍Assistance in multitasks or managing multiple tasks: current productivity tools only let you enter multiple tasks, and/or describe the relationship between the tasks, but they do not assist in the preparation, execution, or management of those tasks.
• ‍Assistance with tackling large projects to make them more manageable - there are no tools on the market that assist the user in breaking down large tasks or projects into smaller, more manageable ones.
• ‍Proactive Assistance - In order to be their most productive selves, users want an assistant that acts more like an executive assistant—one that deeply understands the user and can anticipate their needs by preemptively offering assistance, solutions, and suggestions.
• ‍Focus and Accountability - Users need an environment that sets them up for success. Removing distractions and holding them accountable for their commitments, motivating them along the way.

‍

‍

[[symbol=divider]]

User Personas & Customer Journey

Through comprehensive user interviews, we identified several primary customer segments both with and without ADHD: young adults, busy professionals, students, and parents. For our Minimum Viable Product (MVP) and alpha release, we chose to focus on busy professionals and students with ADHD. This decision allows us to address a clear and impactful problem, ensuring our solution meets the specific needs of a highly affected group.

Our development rationale is straightforward: if our assistant effectively helps the ADHD community stay focused and on task, it will undoubtedly provide value to the broader non-ADHD market. The challenges faced by neurodivergent individuals often overlap significantly with those experienced by extremely busy professionals and overstimulated students. These groups are overwhelmed by numerous responsibilities, long working hours, tight deadlines, and constant demands—leaving them stressed with little time for personal passions, self-care, or family.

There is a clear need for a tool that integrates seamlessly into their lives—much like an executive assistant—to handle administrative tasks and reduce cognitive load. By alleviating routine busywork, users are free to focus on more meaningful and productive activities. Targeting the most overwhelmed users first provides a definitive indicator of our app’s effectiveness in delivering real-time, personalized support. It also serves as the ideal platform to validate our core features before expanding to other user groups in the future.

To gain a deeper understanding of our users’ goals, needs, experiences, and behaviors, I developed detailed user personas and customer journey maps. These tools have been instrumental in aligning our product development with user expectations and ensuring that we create a solution that truly resonates with our target audience.

‍

[[symbol=user-persona-spicybrain]]

‍

To create the journey map, I then whittle the persona down even further to two core user attributes:

• User Needs
  • The top one or two solutions the user absolutely needs to accomplish their goals
• User Mindset
  • One or two sentences that clearly communicates the mindset with which the user will approach our app. It is through this lens that I will look through to see the product as the user would.

This will be the lens through which we view the user journey map.User JourneyBecause AI productivity apps are still relatively new, much of the data for the typical user experience comes from standard productivity apps.  The AI user journey requires careful evaluation and a vastly different approach, as AI products simultaneously solve problems differently, and also solve different problems.Key Differences Between AI and Standard Apps

• AI apps offer personalized, adaptive experiences that evolve based on user behavior, while standard apps provide static, rule-based functionalities⁠⁠
• AI apps require more extensive onboarding to explain adaptive features, whereas standard apps have straightforward, feature-focused onboarding⁠⁠
• AI apps continuously improve through machine learning, while standard apps remain consistent between updates⁠⁠

Challenges and Considerations for AI Apps

• AI apps face challenges in managing user expectations and building trust due to their unpredictable nature⁠⁠
• Data privacy and ethical considerations are more significant for AI apps due to their reliance on extensive user data⁠
• AI apps must balance personalization with transparency to maintain user trust and engagement⁠⁠

Enhancing User Experience in AI AppsNatural Language Processing (NLP) enables more intuitive, conversational interactions in AI apps, improving accessibility and engagement⁠Successful AI apps incorporate user feedback mechanisms and clear explanations of AI decision-making to build trust and improve functionality. For a more detailed evaluation of this subject, see my article  AI App User Journey vs Standard Apps.

[[symbol=divider]]

AI Architecture

An overview of the AI Models and the architecture necessary to deliver an assistant capable of understanding a user's goals, interests, priorities and productivity windows. This understanding will act as a foundational criteria for which we organize the user's daily activities such that the user is more productive with SpicyBrain's version of their schedule.

‍

[[symbol=blog-info-cards, id=ai-models]]

‍

[[symbol=divider]]

‍

Data Strategy - AI Input & AI Output

The most critical step in planning your AI application is your Data Strategy. This is where you decide what data your models need to accurately perform their job, how you will acquire that data, how you will process and output that data, how you will measure the success of the output for that data, and what mechanisms you will deploy to improve upon those metrics. This is truly the heart of any successful AI product.

‍

[[symbol=blog-info-cards, id=ai-input-output]]

‍

[[symbol=divider]]

‍

Data Pipeline

Step 1 - Data Collection

• Gather user input (tasks, goals, interests) via app forms and voice commands.
• Sync with external calendars (Google Calendar, Outlook) for schedules.
• Track task completion and behavior patterns in-app.
• Integrate wearable data (optional) to monitor energy or focus levels.

‍Step 2 - Data Preprocessing

• Clean user input to ensure consistency (e.g., format dates and task labels).
• Deduplicate recurring events and tasks from multiple sources.
• Map tasks to predefined categories (e.g., work, personal, fitness).

‍

Step 3 - AI Model Preprocessing

• User Embeddings: Convert preferences and goals into vector representations.
• Transformer Models (NLU): Tokenize and encode natural language input.
• Seq2Seq (Task Decomposition): Translate complex tasks into actionable steps.
• Reinforcement Learning: Process user feedback for future recommendations.

‍

Step 4 - Output Generation and Feedback Loop

• Generate personalized, prioritized task lists in real-time.
• Suggest focus window adjustments based on energy patterns.
• Present task decompositions for larger projects.
• Collect user feedback on recommendations to continuously refine outputs.
• Adapt priorities dynamically as user schedules evolve.

‍