Your SLA. Your API docs.
The system builds the flows.
You approve. It runs.
Upload your contracts and your booking system's API request/response samples. The system reads both, generates operational call flows autonomously, presents them for your approval, and executes them across voice and email — tied to one case per booking.
Most travel AI is a chatbot that deflects tickets.
It answers FAQ. It opens a ticket. It hands off to a human for anything real. It does not read your contracts. It does not know your fare rules. It does not touch your booking system.
Answers FAQ. Deflects tickets. No operational logic.
Does not read your SLA. Does not touch your booking system.
An operations layer that runs your contact center.
The system reads your SLA and your API docs, generates operational flows autonomously, and executes them across voice and email — with every interaction tied to one case per booking.
Reads your SLA + API docs. Builds operational logic without human coding.
Voice, email, and API data — one case per booking, not per channel.
Handles schedule changes, rebooking, refunds, retention — live on calls.
Same flows as voice. Inbound email to outbound reply, end-to-end.
From contract to live call. Autonomously.
Upload
Your SLA and your API request/response samples. The system consumes both inputs simultaneously.
Generate
The system autonomously bridges contract clauses to API calls — building operational flows without manual coding.
Approve
Review every generated flow. Accept, reject, or modify. Your approval gates what goes live.
Execute
Approved flows become live rules. The agent handles voice calls and email — one case per booking.
The system does not use templates. It reads your actual SLA and your actual API docs, then generates flows specific to your operational logic. When your contracts change, the diff is auditable.
One case per booking. Voice + email + API data.
Every interaction about the same PNR lives in one case. The agent sees the full history — regardless of channel.
One pane for every call. Every email. Every API call.
- VOICEin progress02:34Schedule change · LHR → NRT · PNR XK7-92Fcase 2026-04-9214 · locale: ja · flow: schedule_change_v3 · API: /rebook
- EMAILreview00:38Refund request · booking BR-44188case 2026-04-9213 · auto-replied · clause matched · API: /refund
- VOICEresolved04:12Cancellation · PNR ZQ2-118case 2026-04-9201 · flow: cancellation_v2 · 30% retention · API: /cancel
- EMAILresolved00:04Itinerary confirmation · BR-44102case 2026-04-9198 · auto-sent · locale: de · API: /confirm
- VOICEescalated06:48Group booking modification · PNR HM4-771case 2026-04-9192 · escalated · API: /modify queued
Latency is the only number a buyer can hold us to. The rest tells you what the console tracks.
Reproducible on the demo call.
Not roadmap. What you can measure today on our production stack.
SLA + API docs → operational flows
Inbound to outbound. Same flow as voice.
p50. Human pause is 200-500ms.
Detected at call open. Switched mid-turn.
Dogfooded before it reaches you
- FlexoTripsvoice + emailOur metasearch travel agency — real bookings, real ops load.
- AiOrchemail onlyThe multi-agent code orchestrator we use to ship TravelCalls.ai itself.
Detect the language. Route the call. Hand off mid-turn.
Each call opens with detection. When confidence crosses threshold, the agent re-binds to that locale — without restarting. Email inherits inbound language.
Send us your SLA and your API docs.
We generate flows in under an hour.
Email the contract and a sample API request/response from your booking system. The system reads both, drafts the operational flows, and surfaces them on a 30-minute call — together we review what it generated, what it got wrong, and what to harden before going live.
The same upload-and-walkthrough becomes your production onboarding. Nothing is staged for the demo.