Intent and trip context
Identify whether the traveller needs status, policy, cancellation, transfer, accommodation or itinerary support.
autoDX developed an AI support assistant that answers policy and booking-status questions, retrieves live itinerary details, performs approved changes through tool calling and hands complex cases to a human with the full context intact.
Talk to our team→A delayed flight, missed transfer or policy question could not wait until the next morning.
Phone and email queues were effective when staff were available, but repetitive after-hours requests accumulated quickly. Many questions also depended on live booking information, so a generic chatbot would not have been enough.
Phone or email becomes the entry point.
The case remains in queue until coverage is available.
Booking, policy and previous messages are checked manually.
The agent makes the change in the reservation system.
Another agent may need to rebuild the same context later.
The assistant works across WhatsApp and web chat, retrieves approved knowledge and booking state, uses controlled tools for permitted actions and moves complex cases into the existing helpdesk.
Identify whether the traveller needs status, policy, cancellation, transfer, accommodation or itinerary support.
Build answers from approved support content instead of relying on model memory.
Read the reservation system before answering booking-specific questions.
Expose only approved write operations with validation and traveller confirmation.
Keep conversation and itinerary context independent from the chat channel.
Create a helpdesk case with summary, transcript, booking context, tools used and unresolved decision.
Cancellation, change and supplier rules evolve and cannot be guessed.
Reading a booking is low risk; modifying it requires tighter control.
A timeout can make it unclear whether a change succeeded.
Forwarding the last message alone does not explain the journey.
| Layer | Technology | Role |
|---|---|---|
| AI orchestration | LLM with tool calling | Intent handling, grounded responses, tool selection and summarization. |
| Booking integration | Reservation / booking APIs | Retrieve itinerary state and perform approved changes. |
| Customer channels | WhatsApp and web chat | Provide a consistent support entry point. |
| Knowledge | Approved policy and support content | Ground case-specific responses. |
| Human support | Helpdesk integration | Receive escalated cases with context and action history. |
| Reliability | Session store, queue and audit logs | Preserve state, manage retries and trace actions. |
autoDX connects AI assistants with booking systems, policy knowledge, digital channels and helpdesk workflows to deliver practical after-hours support.