AI Assistant for a Service Support Team

Task

The support team receives a large daily flow of requests from different types of clients:

• Investors — place investments through the platform.
• Borrowers — receive loans.
• Agents — bring in new clients.

Before implementing AI, all requests were handled by managers, which led to overload, wasted time on spam, and slower responses. The goal was to automate the first line of support and reduce the team’s workload.

Solution

An intelligent chatbot and email assistant were developed and integrated with the ticketing system:

• Spam filtering
• AI is connected not only to the messenger but also to the support email inbox.
• All incoming messages are first checked for spam.
• If identified as spam, the ticket is automatically closed so managers don’t waste time on it.
• Client and topic identification
• The bot analyzes each message and classifies it by two parameters:
• client type (investor, borrower, agent),
• topic category.
• These tags are recorded directly in the ticket system.
• Knowledge-base replies
• A knowledge base was built from the website FAQ and frequently used support responses.
• The bot generates a ready answer and sends it to the client.
• Clear separation of responsibilities with human managers
• If the bot takes a ticket, it marks it as “handled by bot” so managers don’t interfere.
• If a ticket is ambiguous, the bot leaves it untouched and passes it to a manager.
• If a manager has already started working on a ticket, the bot doesn’t interfere either.

Results

• Support team workload was reduced thanks to automated spam filtering and handling of standard requests.
• Client response time dropped significantly since most questions are resolved instantly.
• Managers are focused only on complex cases that require human judgment.
• Ticket system conflicts were eliminated — it’s now always clear whether a ticket is handled by the bot or a manager.