How Bee2Bees DMC answers travel agents in seconds with Claude
Brainam built Bee2Bees DMC a WhatsApp quoting bot and a team CRM dashboard, both running on Claude. Since March 2026 the bot has sent over 29,000 replies, answering in about one second and handling about 9 in 10 customer chats without a human.
At a glance
Bee2Bees DMC, a B2B destination management company for Goa and Himachal
Travel agents across India
Brainam Technologies, New Delhi
A WhatsApp quoting bot and a WhatsApp CRM dashboard
Claude API (Sonnet, Haiku and Opus models), built with Claude Code
March 2026
The challenge
Bee2Bees sells hotels, transport and holiday packages in Goa and Himachal to travel agents, who book them for their own guests. Agents send enquiries on WhatsApp at all hours, in English and Hindi. Each one wants a price fast, often while their guest is still on the line.
Every quote meant reading the details, checking hotel and transport rates, and writing a reply by hand. Then came the changes: "make it 4 star", "change the dates". In peak season, the team spent its day quoting instead of closing bookings, and slow replies risked losing agents to faster competitors.
The solution
Brainam built two connected products on Claude: a WhatsApp bot that quotes, and a dashboard where the team runs everything else.
The WhatsApp quoting bot
The bot runs on Bee2Bees' main WhatsApp business number. A travel agent sends a query in English or Hindi, and the bot:
- reads the trip details: dates, nights, guests, rooms, hotel and meal preferences
- sends back a price quote for hotels, transport or a full package
- updates the quote when the agent asks ("make it 4 star", "change dates")
- answers follow-up questions
- reads payment screenshots and guest ID photos
- hands the chat to a person whenever the agent types "AGENT"
Staff use it too, to request prices from vendors and to set follow-up reminders over WhatsApp.
The team dashboard
The dashboard is Bee2Bees' WhatsApp CRM, with access set per team member. It shows every customer chat from both WhatsApp numbers, and staff can take over from the bot at any point. It also covers:
- a leads pipeline, from new enquiry to booking
- bookings and vouchers
- vendors and the transport ledger
- broadcast campaigns and reminders
- team performance reports
- a knowledge base the bot answers from
- an AI Learning page, now being rolled out, to suggest bot improvements from how the team replies
Why Claude
Fast enough for a live phone call
Agents ask for a price while their guest is still on the line. Claude Haiku handles the high-volume steps, so a typical reply goes out in about a second.
Understands real travel queries
Claude Sonnet reads messy enquiries in English or Hindi and pulls out dates, nights, guests, rooms, hotel and meal preferences.
Reads images, not just text
The bot reads payment screenshots and guest ID photos that agents send on WhatsApp, so the team doesn't have to retype them.
The right model for each job
Sonnet, Haiku and Opus each do the work they are best at, which keeps replies fast and running costs under control.
Brainam also built the product with Claude Code: about 600 of the projects' 840 code commits.
Results
Since going live in March 2026, the bot has answered travel agents in about a second and handled most chats with no one from the team stepping in.
| Result | Figure | How it was measured |
|---|---|---|
| Reply speed | About 1 second; 90% of replies within 6 seconds | 14,341 bot replies |
| Chats handled fully by the bot | About 9 in 10 | 918 of 1,027 customer chats needed no team member |
| Travel enquiries quoted | About 440 a month in season, peaking at 770 in April 2026 | 3,248 quote requests, March–November 2026 |
| Total bot replies | 29,600+ | Replies to 16,893 customer messages since March 2026 |
| Campaign reach | About 21,400 messages, 952 replies (4.4%) | Broadcasts sent from the dashboard |
For comparison, the team's own replies on the human-only WhatsApp line take about 16 minutes for a typical first reply. That figure comes from a small sample of 18 replies.
"Travel agents want a price while their guest is still on the phone. Our WhatsApp bot, built by brAInam on Claude, now replies in about a second in English or Hindi and handles most enquiries on its own. Our team spends its time on bookings instead of typing quotes."
Under the hood
The bot and dashboard run on the Claude API. Each job uses the Claude model that fits it best for speed and cost. Brainam wrote about 600 of the projects' 840 code commits with Claude Code.
| Claude model | What it does for Bee2Bees |
|---|---|
| Claude Sonnet 4.6 | Reads and understands incoming travel queries |
| Claude Haiku 4.5 | Answers questions, applies quote changes, translates replies, reads payment screenshots and guest IDs, parses staff reminders |
| Claude Opus 4.6 | Sorts contacts into categories for campaigns |
| Claude Opus 5.5 | Powers AI Learning (being rolled out), which reviews the team's replies every 6 hours to suggest bot improvements |
Timeline: the bot and the first dashboard version went live in March 2026, with Claude contact sorting from April. A second team WhatsApp line followed in August, and AI Learning in October 2026.
Bee2Bees DMC is a B2B destination management company for Goa and Himachal. It sells hotels, transport and holiday packages to travel agents across India, who book them for their own guests.
Brainam Technologies, New Delhi, builds customised AI solutions for small and medium businesses: AI agents and chatbots, workflow automation and custom AI apps on the tools they already use. brainam.ai
Want AI that answers your customers in seconds?
Tell us how your business handles enquiries today. We'll show you what AI can take off your team's plate.
Talk to us on WhatsApp ← All case studies