Multimodal AI Personas for Diverse Industries

Envision an AI that goes beyond mere responses, engaging in a dialogue that's both familiar and intuitive. Picture delving into local politics with a former president, exploring academic subjects under the guidance of a distinguished professor, or seeking healthcare advice from a doctor who specializes in your condition.

Six Feet Up developed AI personas that are not generalists. The personas are specialists, crafted to meet the unique needs of various sectors including:

• Automotive: AI personas assist customers in navigating catalogs of vehicles, facilitating purchasing or rental decisions through knowledgeable interactions.
• Education: AI personas act as tutors in specific subjects, enhancing learning experiences by interacting with students in a personalized manner.
• IT: AI-driven support systems provide immediate, reliable answers to IT Help Desk queries based on extensive internal manuals and documentation, significantly reducing downtime and enhancing productivity.
• Healthcare: AI personas offer specialized medical advice, catering specifically to individual health profiles and needs, such as advice for women over 40 with particular health conditions.

Using a mix of sophisticated AI tools and technologies, Six Feet Up developed several personas capable of understanding and responding to human queries in real time. This initiative aimed not only to enhance user engagement through personalized experiences but also to streamline operations across multiple industries.

AI Tools

• Speech Recognition and Generation: To create more natural and engaging user experiences, Six Feet Up integrated advanced speech recognition technologies like Deepgram and AWS Transcribe, which convert spoken language into written text. Additionally, the company utilized text-to-speech tools such as Elevenlabs and Coqui TTS, enabling the system to communicate with users in a clear and lifelike voice. This blend of technologies allows the system to handle queries and provide information both by text and speech, significantly enhancing accessibility and user satisfaction.
• Text Processing and Interaction: Leveraging OpenAI’s ChatGPT and open source models including Mistral, Six Feet Up developed a sophisticated text processing system that understands and generates contextually relevant responses. This technology ensures that interactions are not only accurate but also engaging, allowing the system to efficiently handle complex customer inquiries.

Development and Deployment

• Architecture: Six Feet Up adopted an Event-Driven Architecture utilizing Django and MQTT. This architecture supports the dynamic and efficient management of data flows and user interactions across multiple services, improving overall system responsiveness and reliability.
• Development Environment: The deployment strategy included the use of EKS Kubernetes for robust, scalable cloud deployments, ensuring that the infrastructure could handle varying loads with ease. For local development, Tilt was used to streamline processes and enhance developer productivity.
• Real-Time Data Handling: Utilizing websocket technology, Six Feet Up implemented a system for instantaneous updates, crucial for maintaining engagement during user interactions. Additionally, PostgreSQL with pgvector was employed to handle sophisticated data retrieval processes, significantly enhancing the AI’s responsiveness and accuracy.