2 ready-to-build workflows

AI agent workflows for Travel & tourism

Answer itinerary questions, change bookings and collect travel documents — in the traveller's language.

Each recipe below is expressed only in VegaDūta's real workflow building blocks — Trigger, Agent, Knowledge, Condition, Approval, Output, Tool (MCP), Voice, Device, Loop, Code and Parallel — so it maps 1:1 to something you can assemble in the Workflow designer. Consequential actions always pass a human Approval step.

1. Itinerary Q&A and booking changes on WhatsApp

Travellers message the agency at all hours asking what's included, when the transfer leaves, or to move a date — and an agent digs through the itinerary PDF and the booking system by hand, only in office hours.

Trigger: WhatsApp / Call (traveller message)

How the workflow runs

  1. Trigger

    Traveller reaches out. A WhatsApp or Call trigger starts on any inbound message or call, in whatever language the traveller uses.

  2. Knowledge

    Answer from the itinerary. A Knowledge node grounds the reply in the trip's itinerary, inclusions and policy documents — with citations, so nothing is invented.

  3. Agent

    Understand + handle the request. An Agent node interprets the question (what's included, timings, or a change request) and works in the traveller's language via Sarvam.

  4. Condition

    Route change requests. A Condition node branches: simple questions answer straight away; a date/room/booking change continues to the update path.

  5. Approval

    Agent confirms fare-affecting changes. An Approval node holds any change that alters price or availability for a human to confirm — nothing that costs the traveller money moves on its own.

  6. Tool (MCP)

    Update the booking. On approval, a Tool (MCP) node applies the change in your GDS / booking engine. There's no first-party connector — the change is pushed via MCP (or queued to ops).

Channels & connectors

  • WhatsApp
  • Voice / Call
  • Sarvam (22+ languages)
  • Knowledge base
  • Approval / HITL
  • Tool (MCP → booking system)

Outcome

Travellers get accurate, in-language answers instantly and simple changes are made for them — while any fare-affecting change stays human-confirmed.

Why it helps

Replaces the manual itinerary lookup and after-hours silence with an always-on concierge, and keeps the money decision on any change with a person.

Build spec

1 agent — 1 concierge agent handles the conversation; the change-routing Condition, the fare Approval and the booking write are workflow nodes.

System prompt (paste-ready)

You are a travel concierge for {{agency}}. Answer the traveller's questions strictly from the trip's itinerary, inclusions and policy documents in Knowledge, with citations — never invent an inclusion, timing or fee. Work in the traveller's language (Sarvam covers 22+ Indian languages). If the traveller asks to change a date, room or booking, gather the details and prepare the change, but do not apply anything that affects fare or availability yourself — that must be human-confirmed. State clearly that a fare-affecting change is pending confirmation.

MCP connectors

  • GDS / booking engine — via a Tool (MCP) node (no first-party connector)
  • WhatsApp — receive + send channel
  • Sarvam — in-language conversation

Built-in tools

  • knowledge_search (itinerary + inclusions + policy)

Guardrails

  • Grounded-only on the trip's own documents — no invented inclusions, timings or fees
  • No fare- or availability-affecting change without the Approval node — nothing that costs the traveller money moves on its own
  • Scope-locked to the traveller's own booking

Output & delivery

The agent answers in-language grounded in the itinerary → the Condition node routes change requests → fare-affecting changes go to the Approval node → on approval a Tool node applies the change in the GDS/booking engine via MCP (or queues it to ops).

2. Visa & travel-document collection with a completeness gate

Before every departure, a coordinator emails travellers back and forth for passport scans, photos and visa forms, then manually checks each file is present and legible — days of chasing per booking.

Trigger: Webhook (booking confirmed) or WhatsApp

How the workflow runs

  1. Trigger

    Trip is booked. A Webhook trigger (from your booking system) or a WhatsApp message opens document collection for the traveller.

  2. Agent

    Request exactly what's needed. An Agent node asks for the specific documents this destination requires (passport, photo, visa form) and accepts them right in the chat.

  3. Condition

    Completeness gate. A Condition node won't advance the file until every mandatory document is present — incomplete cases loop back with a clear list of what's still missing.

  4. Output

    Pre-trip reminders. An Output node sends timed reminders on WhatsApp/SMS for anything outstanding and the pre-departure checklist as the date nears.

  5. Tool (MCP)

    File the documents. A Tool (MCP) node stores the collected documents against the booking in your system.

Channels & connectors

  • Webhook
  • WhatsApp
  • SMS
  • Sarvam (multilingual)
  • Tool (MCP → booking system)

Outcome

Every traveller's documents arrive complete and on time, checked automatically, without a coordinator chasing email threads.

Why it helps

Removes the days of manual document chasing and the risk of a missing paper surfacing at the airport — the completeness gate does the checking, not a person.

Build spec

1 agent — 1 agent requests and checks the documents; the completeness Condition, the reminders and the filing step are workflow nodes.

System prompt (paste-ready)

You are a travel-document collection assistant. In the traveller's language, ask for exactly the documents this destination requires (per the destination checklist in Knowledge) — passport, photo, visa form — and accept them in the chat, guiding on clarity, format and validity when an image is unusable. Confirm each document as it arrives. Never mark the file complete while any mandatory document is missing or illegible; instead list clearly what is still outstanding.

MCP connectors

  • Booking system / document store — via a Tool (MCP) node (no first-party connector)
  • WhatsApp + SMS — receive + send channels
  • Sarvam — in-language conversation

Built-in tools

  • document parsing (check passport/photo/form legibility)
  • knowledge_search (per-destination required-document list)

Guardrails

  • The file never advances until the Condition completeness gate passes — every mandatory document present and legible
  • Scope-locked to the traveller's own booking and documents
  • PII (passport scans) is handled only within this case and filed to the booking, nowhere else

Output & delivery

The agent requests and collects the documents → the Condition gate holds until all mandatory documents are present and legible → an Output node sends timed reminders for anything outstanding → a Tool node files the documents against the booking via MCP.

Related industries

Build one of these in minutes

The sandbox gives you a live agent workspace — no account, no card. Or head back to the full catalogue and compare patterns across every industry.

Try this in the sandbox — no signupBrowse all industries
See how VegaDūta compares to n8n, Dify and BotpressWhy VegaDūta's architectureEstimate your WhatsApp API costs