The protocol between AI trip planners and hotels.

Strait is the lossless, two-way protocol between AI trip planners and hotels. Modern, rich, and built for agentic scale. Made for agents that book directly and carry language-rich preferences end to end.

Looking for our first design partners.

Become a design partner Read the docs

Personal AI

Claude · ChatGPT · any AI trip planner

strait

The protocol

Every Hotel PMS

Rooms · rates · reservations

Every Hotel AI

Concierge · front desk · ops

Built into every Strait integration.

Agents own the experience

Search, price, book, and cancel all happen inside the agent. No redirects, no handoffs to your competitors.

Search doesn't burn quotas

Cheap, rich discovery up front. Live PMS calls only at booking intent. Your agent can search exhaustively without paying look-to-book penalties.

Preferences travel with the booking

"High floor, dairy-free breakfast." Chat preferences reach the hotel as real instructions, not just checkboxes.

Two-way channel to the hotel

A live link between planners and hotels. Changes and issues surface early, travelers are kept in the loop.

Hotels own the guest

Bookings flow straight to the hotel's own system. No OTA or bedbank in the middle.

The hotel is the Merchant of Record

Payments settle directly to the hotel through its existing processor.

A hotel toolkit for agents.

One install. Your agent gets the tools to search, price, book, and cancel, all inside the chat, without touching a legacy checkout page.

agent.py

# pip install getstrait
from strait import Strait

strait = Strait()
tools = strait.tools.for_anthropic()  # or .for_openai()

# Your agent now has hotel booking tools:
#   find_property       find a hotel by name, vibe, address, or stable place ID
#   check_availability  rooms, pricing, and cancellation policies from the hotel
#   book                checkout card with guest preferences attached
#   cancel_booking      cancel with confirmation

Find the hotel

"Book me the Arlo Soho" or "that boutique hotel near Times Square." Strait's semantic search resolves it to the right property.

hotels.find_property({ name: "Arlo Soho", city: "New York" })

Check availability

Rooms, pricing, and cancellation policies straight from the hotel's PMS. No bedbank, no OTA repack, no third-party rate sheet in between.

hotels.check_availability({ property_id, check_in: "2026-06-10", check_out: "2026-06-12", guests: 2 })

Book the room

A checkout card appears for the guest to review and confirm. Structured preferences still get filled in, and rich preferences from the chat travel with the booking too, reaching the hotel as real instructions instead of being dropped at the API boundary.

hotels.book({ offer_id: "off_8f72b9a1", guest_name: "Jane Doe", preferences: "high floor, away from elevator, dairy-free breakfast, sparkling water in the room" })

Booking confirmed

A real reservation is created in the hotel's system. Guest preferences arrive with it. All without the AI ever seeing keys, cards, or financial data.

Works with your agent.

MCP Server

Connect Claude Desktop, Cursor, Anthropic's API, or any MCP client to mcp.getstrait.dev.

Python SDK

pip install getstrait. For custom agents built from scratch on Anthropic, OpenAI, or any other model.

REST API

Direct HTTP for agent platforms, enterprise stacks, and runtimes that don't speak MCP.

Vercel AI SDK LangChain OpenAI Agents CrewAI LlamaIndex

Become a design partner.

We're working hands-on with our first agent builders and hotel partners to shape the protocol. Leave your email and we'll be in touch.

You'll be emailing Omar, the founder.

No spam. We'll only email you about access.