1 Week
Research & Planning
Studied user behavior, voice accessibility needs, and system requirements for the disability care domain.
This project involved building a generative AI voice assistant tailored for a disability support services company. The client, an NDIS-registered provider, needed an intelligent system to automate inbound inquiries, schedule appointments, and handle frequently asked questions. The primary goal was to improve accessibility for clients with disabilities while reducing the workload on human staff. The assistant was integrated with voice recognition and built with customized workflows for support service delivery.
Accessibility Improved: Enabled users with disabilities to get information and schedule services using voice alone.
Staff Workload Reduced: The AI assistant handled over 70% of incoming inquiries, freeing staff for higher-value tasks.
Response Time Enhanced: Immediate and accurate responses reduced average client wait time by half.
Scalability Achieved: The solution is scalable and can support multiple support centers simultaneously.
Positive User Feedback: Clients praised the convenience and speed of accessing services through voice.
The main technical challenge was building an AI voice assistant that could understand diverse accents and speech impairments while maintaining high accuracy. Integrating seamless voice-to-text and text-to-voice interactions using OpenAI and Whisper APIs also required optimization. Ensuring secure handling of sensitive health-related information under NDIS compliance was a crucial part of the project.
1 Week
Studied user behavior, voice accessibility needs, and system requirements for the disability care domain.
3 Weeks
Developed and trained the AI voice assistant with natural language understanding and intent recognition.
2 Weeks
Designed and integrated custom support workflows for call routing, appointment scheduling, and service intake.
1 Week
Conducted end-to-end testing with real user scenarios and deployed the system with logging and monitoring.
Analyzing customer behavior on the web pages, predicting and preventing revenue losses.
Talk to her and stop being lonely.
Multilevel hierarchical knowledge and memories are built with LangChain.