WLLNSS AI Assistant
Intelligent conversational assistant integrated into the WLLNSS platform to boost employee engagement with wellness services, benefits, and healthy lifestyle practices.
Client:
WLLNSS, Europe
Project Overview
WLLNSS partnered with Axis to create an AI assistant that goes beyond a simple chatbot. The goal was to help employees better understand and use the wellness services available to them, while also encouraging healthier everyday habits.
The assistant acts as a personal wellness guide, capable of holding natural, supportive conversations, recommending relevant services, and launching structured scenarios such as challenges or surveys. It integrates with company policies, insurance details, calendars, and wearable devices to provide timely, personalized, and actionable suggestions.
A key requirement was ensuring the system worked across different client companies, each with unique benefit packages and rules. At the same time, it needed to maintain strict safety boundaries, avoiding medical advice but offering clear escalation paths for urgent cases.
Challenge
The client required an assistant that was not only conversational but also action-oriented and safe. Core challenges included:
Personalization at Scale: Tailoring recommendations based on user data (surveys, wearables, schedules) while ensuring strict compliance with corporate policies.
Boundaries & Safety: Enforcing guardrails to prevent medical advice, while providing escalation protocols for critical “red flag” cases.
Multi-Company Adaptability: Ensuring the knowledge base and policy rules could flexibly adapt to different corporate clients and insurers.
Engagement & Conversion: Moving beyond static FAQs to encourage action—triggering surveys, launching widgets, and nudging lifestyle improvements.
Tech Stack
LLM Provider: Google Gemini (multilingual embeddings and text generation).
Databases: Milvus vector database for policy, insurance, and FAQ retrieval.
Architecture: Orchestrator for intent detection, guardrails, and integration with backend services.
Channels: WLLNSS web & mobile apps, with RU/EN support (multi-language expansion planned).
Privacy: Data minimization, pseudonymized logs, no PII stored by the assistant.
Solution
Axis implemented a retrieval-augmented, multi-chat AI assistant that combines proactive guidance with structured navigation:
Benefits Navigator & Q&A (RAG): Answers user questions by pulling relevant policies, FAQs, and insurance details from the vector database.
Contextual Recommendations: Dynamically suggests wellness actions (e.g., reminders, meditation, surveys) based on user inputs and wearable data.
Scenario & Widget Launch: Identifies intent and offers structured, step-by-step mini-flows that directly launch platform widgets or services.
Proactive Surveys: Engages users with timely surveys and explains their purpose while respecting data privacy rules.
Multi-Chat Architecture: Branches into sub-chats for focused problem-solving (e.g., sleep program, wellness challenge) while maintaining progress tracking.
Safety Guardrails: Blocks medical diagnoses, routes emergencies to professional support, and enables warm handover to HR or telemedicine providers.
The assistant now acts as a wellness engagement hub, turning everyday interactions into opportunities for healthier behavior and increased benefit utilization—measured by usage growth, scenario completion rates, and contribution to reduced absenteeism.
