The Feminine Intelligence Agency (FIA) has developed a research-backed web app called ChatBoy, designed to help users recognize and respond to manipulative communication. The app includes over 1,700 scenario-based questions and responses, each carefully crafted and approved by psychologists using a behavioral trait scoring system built over years of research.

Right now, ChatBoy functions like a quiz—it’s effective, but static. This project invites a student to help us build the next-gen MVP by transforming those same questions and characters into interactive, personality-driven AI agents using generative AI tools.

We’re not looking for a polished or commercial product—just something functional that demonstrates how ChatBoy could evolve. We’ll provide all the structured content: character personalities, speech patterns, and behavioral logic. Your job is to plug that into an AI architecture (like LangChain, GPT Agents, or something similar) to produce 16 dynamic character bots that users can talk to.

You’ll take our data, add some flair, and create an MVP that shows what’s possible when real-world behavioral insights meet modern AI.

Student Deliverables

Agent Design Framework

• A short technical document (~2–3 pages) that maps FIA’s existing character traits and behavioral logic to AI agent architecture.
• Demonstrates how personalities (e.g., Domineering Devin, Egocentric Evan) will be translated into prompts, behavior rules, and agent memory/state. (FIA will provide all the research behind this, so the student need not have any psychological knowledge)

MVP ChatBot Prototype

• A functional web-based or local chatbot interface where users can select from at least 8–16 distinct AI agents (FIA will provide profiles).
• Each agent should simulate a unique manipulative communication style, using pre-approved dialogue from FIA’s 1,700-question database.
• Agents do not need to “think” in real time—pre-scripted logic, persona anchoring, and memory-based interactions are acceptable.
• MVP must be demo-ready, even if not commercially optimized (e.g., no need for styling or production-level deployment).

Agent Personality Deployment

• Each character (e.g., Egocentric Evan, Needy Noel, etc.) must be mapped to their communication style and loaded into the MVP system.
• The student can use copy-paste or templated logic from FIA's existing CSV/personality matrix to scale up across agents efficiently.

Agent Instruction Prompts + Behavior Library

• A reusable prompt framework or configuration file that can be adjusted for new characters in the future.
• A library of 16 characters with prompt instructions, tone anchoring, and dialogue examples already loaded and tested.

Documentation for Handoff

• Technical handoff guide explaining:
• System architecture and how the agents were deployed
• How to swap in new characters or update conversation logic
• Recommendations for next steps (e.g., adding emotional awareness, better memory retention, or integrating psychologist review flow)
• 
  

Demonstration Video

• A short (~3 min) screen-recorded walkthrough of how the chatbot works, showcasing at least two agents and their contrasting styles.