AI travel agents
A trip-planning agent (LangChain, AgentKit, etc.) books a flight + hotel — but the user lands abroad and the device they use to receive boarding passes has no data.
Add a `provision_travel_data` tool to the agent's toolbelt. The agent runs it as part of trip prep, pays from a budgeted wallet, attaches the QR to the trip artifact. User opens the trip card, scans the QR before boarding.
# Excerpt from a trip-planning agent's tool definition
@tool
def provision_travel_data(country: str, days: int) -> dict:
"""Buy mobile data for the user's upcoming trip."""
plan = pick_plan(country=country, min_days=days)
r = httpx.post(X402_API + "/order", json={"plan_id": plan.id})
# 402 — pay from agent's budget wallet
pay(to=r.headers["x-payto"],
amount=r.headers["x-amount"],
chain=r.headers["x-chain"])
esim = poll_until_delivered(r.json()["order_id"])
return {
"qr_image_url": esim["qr_image_url"],
"activation_link": esim["activation_link"],
"expires_in_days": days,
}- ▸Agent budgets are typically $5-50/trip — well within the eSIM data price range
- ▸No need for human-in-the-loop approval (the budget IS the approval)
- ▸eSIM activates only on first connection in destination — agent can pre-buy weeks ahead