Mentalityy

Project Purpose:
Many people experience situations where they need to talk to someone—whether to unload emotional burdens or seek trusted advice. However, not everyone has consistent access to a trusted confidant. Data suggests this is a widespread issue.

Let the Data Speak:

• Fig 1: Current generation reports the highest levels of loneliness (79%)
  
• Fig 2: Number of close friendships significantly declined from 1990 to 2021
  
• Fig 3: The most alarming trend: the number of suicides has been rising since the 2000s
  

Project Motivation:
This project was motivated by the common challenges individuals face when building and maintaining social connections, particularly in unfamiliar environments. Many people experience isolation, limited access to affordable mental health support, and long waitlists for institutional resources. In such contexts, conversational AI tools are often used as an accessible outlet for emotional expression and reflection. These tools can help users externalize thoughts, reduce emotional overload, and reframe situations more realistically.

However, several limitations were observed. Conversations often lacked continuity, requiring users to repeatedly reintroduce personal context. Early responses tended to be generic until sufficient background was provided. Despite these limitations, conversational AI interactions may still reduce immediate stress and prevent the accumulation of anxiety over time.

Conceptual Key Issues:

1. Lack of long-term memory.
2. Lack of short-term relevance.
3. Its communication style being more "chat-botish" instead of "humanish."
4. It never checked in on the user.

These four solutions are crucial to make it feel less like a generic chatbot and more like a caring friend.

Technical Key Features:

1. Long-Term Memory:
   Dynamic Categorical Memory: After each interaction, tailored GPT agents update a set of summaries related to aspects of life:

       core_values_and_beliefs, mental_and_emotional_well_being, family_relationships, health_issues, personal_background,
       aspirations_and_fears, profile_summary, strengths_and_weaknesses, romantic_relationships, social_circle_dynamics,
       daily_routines, work_environment_and_dynamics, most_recent_challenges, most_recent_accomplishments,
       emotional_triggers, communication_style, hobbies_and_interests, personal_development_and_skills,
       financial_situation_and_goals, academic_performance_and_experiences, social_life_and_friendships,
       physical_health_and_lifestyle, personal_interests_and_hobbies, past_traumas
       

   The existing summary in each category is updated from the current interaction. Once updated, a master summary is generated, which guides the next interaction.
2. Short-Term Relevance:
   Targeted Retrieval: During conversation, a specialized GPT agent decides whether pulling up any categorical summaries is relevant. It can return an empty list or a list of three most relevant summaries. If not empty, a summarizing agent integrates them into the conversation seamlessly, providing GPT with relevant information.
3. Chatbot Style Improvement:
   Three Layers of Humanization:
   1. Master Prompt: Directs GPT to respond in a human-like, “SMS-chat-with-friend” style.
   2. Personal User Communication Preference: Pulled from the "communication_style" category and dynamically updated after each interaction.
   3. Humanize Filter: This top-layer filter ensures the response style is relevant to the last ten messages, splits long answers into multiple messages, and removes unnecessary periods to emulate human SMS texting.
4. Check-In Feature:
   Check-In Scheduler: After each interaction, this agent evaluates if a check-in message would be beneficial. It generates a check-in message and schedules a follow-up (within 1-24 hours) based on the context. The check-in message is sent to the user at the scheduled time.

Goals:
The COVID vaccine does not guarantee that an individual will never get COVID. Similarly, this bot does not promise to completely eliminate loneliness and isolation. Just as the vaccine lowers the risk of severe complications, the goal of this project is to reduce levels of loneliness and isolation, potentially helping prevent loneliness-related suicides.

Challenges:

• Prompt engineering was a major challenge. The project contains 35 prompts for 10 tailored agents, each requiring careful design to ensure the desired outcomes.
• Achieving a human-like SMS conversation style required multiple layers of abstraction and numerous failed prompt iterations.
• Ensuring that LLM agents consistently returned outputs in expected formats suitable for downstream algorithms was difficult.

Key Accomplishments:

• The system demonstrates a conversational style that resembles a caring friend, achieved through extensive prompt testing and iteration.
• Targeted retrieval dynamically analyzes which memory aspects are relevant to the current conversation, then pulls, summarizes, and injects them contextually.
• Check-in scheduling provides a proactive and supportive feature by analyzing conversations and scheduling follow-ups, similar to professional care practices.

What Was Learned:

• Fundamentals of prompt engineering.
• Designing AI LLM agents with strict output format requirements.
• Coordinating multiple agents to work cohesively: generating summaries, scheduling check-ins, and updating database records.

Future Vectors:

• Send memes to the user.
• Improve targeted retrieval call logic.
• Add scheduled user suggestions for deeper communication style personalization, such as: "How is my conversation style so far? Is it too optimal/dull? I can adjust."
• Check-in scheduling: Improve the time scheduling mechanism.

Demo: