Blackrock AI agents are designed around specific travel jobs, with access to the context and controls required to be useful inside real operations. Built for travel businesses in Saudi Arabia, the UAE and the wider GCC, they work on WhatsApp, in the inbox and inside WebAdmin, and hand over when a person should decide.
Sample screens and conversations. Travellers, agencies and bookings are fictional.
Agents work from your live supplier data: 200+ supplier connections across the travel ecosystem
Blackrock AI agents are designed around those specific jobs, with access to the context and controls required to be useful inside real operations.
Each agent has a specific job, done within your own rules, and hands over when a person should decide.
The WhatsApp agent answers travellers in the chat they already have open: it understands the request, prices it from your connected suppliers and sends the itinerary as a PDF.
The email agent turns a short, messy email into priced options, a policy verdict and an approval request. Then it books, once the right person says yes.
WebAdmin Copilot watches bookings, suppliers and payments, then proposes the fix. It works from the same live records your team sees.
We will show you what an agent could take on.
Beyond WhatsApp, email and WebAdmin, Blackrock AI takes on defined jobs in CRM, operations, customer support and intelligence. Each works from the same live records and is built to be reviewed by a person.
The CRM Agent can bring together recent commercial activity and available customer context to help teams prepare and follow up faster.
Explore CRM intelligenceThe Operations Agent can review workflow states and operational signals to surface items that need human attention.
Explore AI operationsThe Customer Agent assists support teams and, where appropriate, travellers or agents with questions about the platform and the booking journey.
Explore AI customer supportThe Intelligence Agent lets authorised teams explore operational and commercial data through natural-language queries and generated summaries.
Each agent has a defined job and works from the same live booking, client and supplier data your team sees. Permissions follow the user, and a person stays in charge.
Your records
Scoped to the user's permissions. High-impact actions can require confirmation.
Where people meet them
Agents hand over when a person should decide. Humans approve anything that moves money or messages a client.
We begin by identifying the repetitive decision or information problem, then design the agent around the data, tools and controls required to solve it.
Pick the repetitive decision or information problem that costs your team time.
Decide what the agent can see and which tools it can use: your bookings, clients, suppliers, tasks and conversations, scoped to the task.
Agree what needs a person, from confirmation on high-impact actions to the activity log your team reviews.
Humans approve anything that moves money or messages a client. Bookings outside policy wait for the approver, payments, rebookings and client messages from the Copilot wait for your approval, and a person takes over WhatsApp conversations when the traveller asks or a decision is needed.
Agent permissions reflect user permissions, and sensitive data access is scoped to the task. Agents work from the same live booking, client and supplier records your team sees.
Agent activity is logged where appropriate, and approvals sit where the outcome matters, so your team can review what happened.
Start with the workflow, not the model. We identify the repetitive decision or information problem in your operation, then design the agent around the data, tools and controls needed to solve it.
The Customer Agent supports English and Arabic workflows, and it escalates to a person when confidence or permissions are insufficient.
A separate chatbot is easy to demo and easy to ignore. Blackrock AI connects agents to real business context (customers, bookings, tasks, conversations, opportunities and operational events) inside the workflows your team already uses.