1to1Help
India
Healthcare & Wellness
Amazon Bedrock, Amazon S3, AWS India Regions, Vector Database, FastAPI, Streamlit
1to1Help
India
Healthcare & Wellness
Amazon Bedrock, Amazon S3, AWS India Regions, Vector Database, FastAPI, Streamlit
1to1Help, India’s largest Employee Assistance Program provider, needed to scale how it supports employees seeking emotional wellness resources. Their existing chatbot flow depended on manual triage and human-driven scheduling, creating bottlenecks and limiting growth. Avahi implemented an agentic Al system that automates query handling, wellness activity recommendations, counselor matching, and scheduling, while routing sensitive cases to humans for oversight. The solution increased operational scalability and maintained strict data residency by keeping all processing within AWS India regions.
1to1Help is a healthcare and wellness services provider that delivers Employee Assistance Programs to organizations across India. Its platform offers one-on-one counseling, digital wellness tools, and group wellness activities, supporting a large population of users seeking mental health guidance and resources.
As usage increased, 1to1Help’s chatbot workflow became constrained by manual chat triage and human-led scheduling steps. Routine requests required unnecessary human effort, slowing response times and reducing the platform’s ability to scale efficiently.
In addition, the platform needed a reliable way to detect out-of-scope requests and identify sensitive or threat-to-life scenarios that required immediate human intervention. Another key challenge was accurately recommending relevant group wellness activities and matching users with appropriate counselors based on specialization and availability, then seamlessly booking sessions. All of this had to comply with data residency requirements, with processing restricted to AWS India regions.
AWS enabled 1to1Help to meet strict India data residency requirements by keeping data processing within AWS India regions. It also provided the foundation to implement secure storage, scalable backend services, and generative Al capabilities for natural language understanding and orchestration.
Amazon Bedrock provided access to large language models for intent detection and natural language processing, supporting an architecture that could automate routine interactions while preserving human oversight for sensitive situations.
1to1Help chose Avahi for its ability to design and implement an agentic Al architecture that balanced automation with safety, especially for mental health and wellness workflows. Avahi delivered an end-to-end solution that automated triage, recommendations, matching, and scheduling, while ensuring sensitive cases were flagged for human review. Avahi also ensured the system complied with India data residency requirements by restricting processing to AWS India regions.
Avahi built an agentic Al system to automate and orchestrate multi-step employee wellness interactions, from initial user message to recommendations and booking. At the front of the workflow, a triage agent performs keyword and intent classification to understand the user’s request, detect unsupported request types, and identify out-of-scope queries.
For safety and oversight, the system includes an automated flagging mechanism that escalates sensitive and threat-to-life scenarios to humans. This ensures routine cases are handled efficiently while high-risk cases receive appropriate intervention.
To recommend relevant group wellness activities, Avahi implemented an activity matching agent using hybrid retrieval that combines semantic similarity (vector search) with lexical matching. This improves recommendation relevance across varied user phrasing and intent. The recommendation engine returns 5 to 8 wellness activities tailored to the user’s input.
For care continuity, the system matches users to counselors based on specialization and availability, then integrates directly with 1to1Help’s scheduling API to complete session booking. A FastAPI backend provides a RESTful orchestration layer, and a Streamlit UI supports testing and validation. All data processing and system operations were restricted to AWS India regions to meet residency requirements.
The solution automated routine wellness interactions while preserving human oversight for sensitive cases, improving scalability without compromising care standards. 1to1Help gained a more efficient workflow for triage, recommendations, counselor matching, and scheduling, with data residency maintained in AWS India regions.
Let’s explore your high-impact AI opportunities together in a complimentary session
Let’s explore your high-impact AI opportunities together in a complimentary session
Let’s explore your high-impact AI opportunities together in a complimentary half-day session
