PsyKinator

PsyKinator is an AI tool that helps to understand your mental state using Akinator style adaptive questioning and visual mind mappinggiving clarity without self-diagnosis and guiding your next steps.

Comment

PsyKinator

Inspiration

This project was inspired by a close friend’s recent diagnosis of depression.

What stood out wasn’t just the diagnosis, but what came after. Even with medication, they still struggled to understand what was actually working and what wasn’t. Some days improved, others didn’t, and there was no clear way to interpret those fluctuations.

At the same time, self-diagnosis has become extremely common. People often label themselves with conditions like depression or anxiety without understanding the underlying causes, whether it’s burnout, chronic stress, or physiological factors like elevated cortisol.

I realized something important:

$$ \text{Clarity} > \text{Labels} $$

So I built PsyKinator to help users understand their mental state and decide what to do next, without pretending to diagnose them.

What it does

PsyKinator is an interactive mental state exploration tool inspired by an Akinator-style questioning system.

Adaptive 20-Question Engine

  • Users answer a dynamically generated sequence of questions
  • Each answer narrows down possible mental states

  • The system outputs:

    • likely patterns (burnout, stress loops, depressive symptoms)
    • suggested next steps

The narrowing process can be modeled as:

$$ S_{t+1} = S_t \cap Q_t $$

Where:

  • $S_t$ is the set of possible mental states at step $t$
  • $Q_t$ represents the constraint introduced by the user’s answer

Mental State Mind Map

  • User responses are transformed into a visual graph
  • Nodes represent emotions, behaviors, and triggers
  • Edges represent relationships between them

$$ G = (V, E) $$

Where:

  • $V$ represents mental state components
  • $E$ represents relationships between them

This allows users to see their mental patterns instead of guessing them.

Tracking and Reflection

  • Users can log sessions over time
  • Compare how their mental state evolves
  • Identify patterns such as improvement, stagnation, or worsening

Guidance, Not Diagnosis

PsyKinator does not diagnose.

Instead, it helps users:

  • understand what they may be experiencing
  • determine whether further attention is needed
  • take small, actionable steps

How I built it

I built PsyKinator as a solo developer using a fast, AI-assisted workflow while maintaining full control over the system design and logic.

Tech Stack

  • Frontend: React-based web application
  • Backend: Lightweight database for storing sessions and state history
  • AI Layer: Structured prompting for reasoning, question generation, and pattern detection

Core Systems

Question Engine

  • Dynamically adapts based on previous answers
  • Mimics Akinator-style narrowing logic

State Interpretation Layer

  • Converts user input into structured signals:

    • emotional indicators
    • behavioral patterns

Mind Map Generator

  • Builds a graph representation of the user’s mental state
  • Updates continuously as new data is added

Challenges I ran into

Avoiding “Fake Diagnosis”

It was easy to accidentally build something that sounded like a diagnostic tool.

I had to carefully design outputs to:

  • avoid labeling users
  • avoid medical claims
  • still provide meaningful insights

Making AI Actually Useful

Generic AI responses are not helpful.

I focused on:

  • structured outputs
  • actionable suggestions
  • avoiding vague or repetitive responses

Modeling Mental States

Human emotions are complex and non-linear.

Representing them as structured data without oversimplifying was one of the hardest parts of the project.

Speed vs Understanding

Using AI tools allowed rapid development, but I made sure to:

  • understand every component
  • maintain full ownership of architecture decisions

Accomplishments that I'm proud of

  • Built a mental state mapping system instead of a generic chatbot
  • Avoided harmful self-diagnosis while still providing useful guidance
  • Created a visual representation of mental states
  • Delivered a full-stack working product as a solo developer within hackathon constraints
  • Designed a dynamic, adaptive questioning system

What I learned

  • Clarity is more valuable than complexity
  • AI is only effective when structured properly
  • Ethical constraints improve the product
  • The best tools guide users rather than replace decision-making

What's next for PsyKinator

Short Term

  • Improve accuracy of mental state detection
  • Refine question flow
  • Improve UI/UX for clarity and calmness

Long Term

  • Personalized long-term pattern tracking

  • Integration with:

    • sleep data
    • activity levels

  • Smarter intervention suggestions

Ethical Expansion

  • Stronger safeguards for:

    • crisis detection
    • escalation to human help

  • Clear communication of limitations

PsyKinator does not try to tell users what is wrong with them. It helps them understand themselves well enough to figure that out.

Share this project:

Updates