AI Chatbot for Personalized Care

Six Feet Up developed a customized prototype, carefully balancing security, AI sophistication, and user experience.

Here’s how it was done:

Backend Development and API Integration

• Six Feet Up built a secure and flexible backend using Django and Django-Ninja which connected seamlessly with the client’s mobile frontend. The API orchestrated all interactions between the app and the AI chatbot, ensuring smooth, real-time communication. Through automatically generated documentation, the client’s team gained a clear understanding of the API, making it easy to maintain and adapt as needed.

Chatbot Prototype and LLM Integration

• Six Feet Up developed a chatbot prototype using OpenAI’s language models and LiteLLM with a Pre-Generated Answers (PGA) approach, enabling expert-approved, context-specific responses that ensure accuracy and reliability. Designed to handle complex queries, PGA draws from a curated Q&A database, where user queries are matched to verified answers, providing consistent responses that build trust — a critical need in a regulated field like healthcare. By incorporating past user interactions, the AI delivers natural, conversational responses, and its scalable design allows the Q&A library to expand as new questions arise, without requiring costly retraining.

Scalable Architecture‍

• To deliver quick and accurate responses, Six Feet Up utilized cloudnativepg and Postgres with pgvector, optimizing data retrieval and question-matching within the chatbot’s database. Scaf™, a project bootstrap tool designed by Six Feet Up to streamline development and deployment processes, was used to deploy Kubernetes clusters for both local and production environments. This architecture supports, fast, scalable, and secure data handling.

Data Security and Compliance

• Ensuring HIPAA compliance was essential. Six Feet Up implemented Redis for secure data storage, enabling encrypted caching for all personal data so only authorized users could access sensitive information. Each session was safeguarded with the client’s mobile authorization mechanism and encrypted user ID, ensuring robust privacy for every interaction.

Enhanced User Experience and Tracking

• The system tracked all user interactions, identifying any unanswered questions and logging user history, allowing for a seamless conversation flow even across different sessions. Six Feet Up also implemented a risk-word detection feature, enabling the chatbot to prioritize questions requiring urgent professional attention. This thoughtful design ensured that the AI was both helpful and sensitive to users’ needs.

Continuous Improvement through Expert Feedback

• Six Feet Up embedded a review mechanism that allowed medical experts to evaluate, update, and approve chatbot responses, enabling the knowledge base to grow and adapt over time. This feedback loop ensures the platform remains accurate and relevant, providing an adaptive and engaging user experience.

Six Feet Up’s team worked hand-in-hand with the client throughout the project, iterating on features, adding LLM-backed responses, and expanding review functionality to fine-tune the system for maximum performance.