She answers, books and resolves on WhatsApp, Instagram, email, web chat and the phone, grounded in your knowledge, calling your own tools, and asking your team before anything that needs a person.
One persona per channel ✦ One tools library ✦ Human approval built in


The most guest-obsessed brands trust Lodgestory to power 50M+ conversations a year
Kaylee is one configuration with three runtimes and a voice. Edit her once and every surface picks up the change on its next run.
Inside a bot journey she answers and resolves end to end: availability, bookings, requests, refunds within policy, on every messaging channel.
Learn moreIn the inbox she drafts the reply, translates the guest and summarises the thread. Your agent edits and sends.
Learn moreAs an AI voice agent she answers calls within a ring, checks your knowledge, calls your tools and transfers to a department when a person is needed.
Learn moreWhen she cannot decide alone, the question or approval goes to the person on call, then the next, then a phone call on your own number.
Learn moreEach AI Agent carries a name, a description and a system prompt of up to 10,000 characters, scoped to one channel. Give WhatsApp a warm concierge, email a measured operations tone and web chat a fast greeter. Scope, tone, rules, what never to do and when to escalate all live in the prompt, so the behaviour is yours to write.
A tool is a named action with a parameter schema in the same function-calling format the major AI platforms use: check availability, look up a booking, fetch the weekend rate, open a ticket, generate a payment link. Attach it to any agent. When a guest asks, Kaylee decides to call it, Lodgestory runs it against your system and she weaves the result into her reply.
checkAvailability({ from: "2026-12-10", to: "2026-12-14", guests: 12 }){ "villas": 3, "rate": 18500, "currency": "INR" }Three Garden Villas are free for 10–14 December at ₹18,500 a night each. Shall I hold them?Upload your policies, FAQs, rate sheets and property guides to a knowledge base and attach it to an agent. Kaylee answers from what you gave her and says when she does not know, instead of inventing a check-out time. Update a document and the next answer uses it.
Reading House policies.pdf · Front desk FAQA fact that is not in her instructions, a policy exception, a price she may not promise: Kaylee puts it to your team as a question, a choice between options, or an approval of a change she proposes. One person is on the hook at a time, the turn passes on its own, the first answer wins, and the customer is told she is checking.
The same AI Agent powers the bot journey that talks to customers, the AI Assistant that drafts for your team in the inbox, and the validation of free-text answers inside bot prompts. Change the prompt or attach a tool and all three use it on their next run. No copying prompts between places, no drift between channels.
Suggested replyOur direct rate this weekend is ₹14,200 a night, below the OTA price. Shall I hold it?Go deeper
Questions, choices and approvals, by turn, with the phone step and ask-back.
Learn moreYour documents, indexed and cited, attached per agent.
Learn moreFunction-calling against your own APIs, with approval where you want it.
Learn moreThe flows Kaylee runs inside, with hand-over, tickets and fallbacks.
Learn moreCustomer stories
Guests message at any hour about arrivals, pickups and late check-outs. Kaylee answers what the knowledge base and the booking say, holds villas through the availability tool, and asks the front desk roster before any refund. When nobody is online at 3 am, she phones the on-call desk on the property's own number and the answer still lands in the guest's thread in her voice.
Every action Kaylee takes is logged with the request, the tool call and the person who approved it. Prompts, tools and knowledge stay inside your organisation, your data is not used to train models, and anything marked Needs approval cannot run without a teammate.
Visit the trust centerWrite one persona, upload one document, attach one tool, and watch her answer a real guest on your own number today.
The things teams ask before they let an AI talk to their guests.
Kaylee is the AI agent inside Lodgestory. She runs inside bot journeys to answer customers end to end, drafts replies for your team in the inbox, validates free-text answers in bot prompts, and answers phone calls as an AI voice agent. Each touchpoint reads the same configuration: a persona with a system prompt, scoped to one channel, and a shared library of tools she may call.
Whatever tools you attach. A tool is a named action with a parameter schema, such as check availability, look up a booking, fetch the weekend rate, create a ticket or generate a payment link. When a guest asks, Kaylee decides to call the tool, Lodgestory runs it against your system and she weaves the result into her reply.
Two ways. The system prompt, up to 10,000 characters, sets scope, tone, rules and escalation triggers. And any change you mark Needs approval, such as a refund, a resolve or a tool call, waits for a person in the Kaylee console. If nobody answers in time it is rejected, not quietly made.
She asks your team instead of guessing. The question goes to the team on call, one person at a time, with a five-minute turn by default; then the next person; then a phone call on your own number when nobody is online. The customer is told she is checking and is never left in silence.
Yes. Each AI Agent is scoped to one channel, so WhatsApp can have a warm concierge, email a measured operations tone and web chat a fast greeter. Tools are organisation-wide, so the same booking lookup works for every persona without duplication.
Yes. The partner API creates and updates AI Agents, tools and attachments; webhooks fire on ai_request.created, answered and replied; and the Lodgestory MCP server lets your own AI assistant answer, approve and ask Kaylee back from outside the CRM.