

Multi-Agent AI System
Multi-Agent AI Booking for Drift Away
Two specialized AI agents work as a single system: they consult the client, pass context between each other, and guide the user from the first question to the completed booking without requiring repeated information.
Technology Stack





Client
A wellness center in Calgary, Canada, offering spa and wellness services, massages, contrast therapy, and other treatments with advance booking available.


Problem
The customer journey to booking involves several different tasks. First, clients need information about services, treatments, prices, and promotions. Once they make a decision, they need to move directly to booking and find an available option.
Additional pressure comes from a high and uneven volume of inquiries to the call center. A significant share of customers ask the same or very similar questions about services, conditions, prices, and promotions, which means the team spends a lot of time on repetitive consultations.
A single universal AI agent quickly becomes too complex in this scenario: it must provide high-quality consultations, navigate a large amount of information, and correctly execute a structured booking process at the same time.
Another challenge appears when the customer moves between different scenarios. If the context is not preserved, users have to explain again which service they selected, which promotion they discussed, and what exactly they want to book.
As a result, the workload on the team increases, while the customer experience becomes slower, less consistent, and less natural.
Solution
For Drift Away, we developed a multi-agent system in which two AI agents have different roles but work within a single customer conversation.
The first agent acts as a virtual concierge. It answers questions about services, treatments, prices, and promotions and helps the client decide what best fits their needs.
When the user is ready to move to booking, the system transfers the conversation to a specialized booking agent together with the context already collected.
For example, if the client first asks about a specific promotion, clarifies what it includes, and then decides to book it, the booking agent already knows which service is being discussed. The user does not need to start the conversation again.
The second agent follows a structured booking scenario, collects the required information, and guides the client through the next stages of the process.
The booking agent does more than simply collect data for a request. It independently checks available time slots, finds sequential and interdependent combinations of several services, including bookings for multiple people, and takes into account the constraints of each treatment.
The system also verifies the client by email: it checks whether the user already exists in the database, sends a verification code, and validates it without staff involvement.
As a result, the entire journey — from consultation and service selection to availability checks, customer verification, and completed booking — can be handled automatically without staff involvement.
For the client, the experience remains one continuous conversation, even though different parts of the process are handled by separate specialized AI agents.

How the System Works
Customer Request → AI Concierge → Consultation & Service Selection → Context Transfer → Booking Agent → Booking

Role of AI in the Solution
AI is used for natural communication with the client, understanding their intent, working with service information, and guiding the conversation according to the current stage of the customer journey.
Each agent has its own role, instructions, and area of responsibility. The concierge focuses on consultation and helping the client choose, while the booking agent focuses on completing the booking process correctly.
A key part of the system is context transfer between agents. This allows the agents to stay specialized without adding extra steps for the user — from the client’s perspective, it remains one continuous conversation.
Result
2 Specialized AI Agents in One System
No Repeated Questions
Dedicated Logic for Each Interaction Stage
Continuous Customer Conversation
Consultation and Booking in One Journey
Automated Customer Service 24/7
Implementation Stages
Weeks 01-02
Design
Analysis of the customer journey, consultation scenarios, booking flow, and interaction logic between the agents.
Weeks 03
Preparation
Structuring information about services, rules, promotions, and preparing the agents’ operating scenarios.
Weeks 04-06
Development
Development of the AI concierge, booking agent, context transfer mechanism, and required integrations.
Weeks 07-08
Testing & Implementation
Validation of real conversation scenarios, context transfer, and the complete customer journey through booking.
Support Period
Evolution
New use cases, scenarios, and functionality.
Support
Conversation monitoring, logic adjustments, response quality control, and system stability monitoring.