In short
- The 2345 hotline handles 20,000 requests a week; the Darobots bot closes 68% of them automatically, with no operator.
- Average response time for a typical question became 4 times faster, operator workload dropped by 40%.
- User satisfaction grew from 61% to 84% in the first two months of the bot's operation.
- The core is AI routing: topic classification, a base of 200+ typical questions, complex cases handed straight to an operator.
About the Project
2345 is a hotline that receives tens of thousands of requests from citizens every week. The topics vary widely: social services, administrative questions, general information. Before automation, operators handled every request manually, which led to queues and slow response times.
The task for Darobots: build a system that takes over routine requests so operators can focus on complex cases.
The Solution
The Darobots team built a multi-tier chatbot with AI-based request routing. The system analyzes each request, determines its category, and either answers automatically or routes it to the right department.
Dialog demo: this is how the bot works in a real chat
Key features:
- automatic classification of requests by topic;
- an FAQ knowledge base with answers to 200+ common questions;
- smart routing: complex cases go straight to an operator;
- request status tracking for users;
- analytics: which questions come up most often, and where the bottlenecks are.
Scale and Technical Challenges
Handling 20,000 requests a week is a different category of reliability requirement. Darobots built an architecture that holds up under peak load without failures.
Results
68% of requests are closed without an operator, and user satisfaction rose from 61% to 84% within the first two months after the bot launched.
Why This Matters for Business
Case 2345 shows that a chatbot with the right architecture can handle a load that would be impossible to cover with people without a major increase in headcount. And response quality doesn't drop - it actually improves, because operators aren't burning out and can focus only on what genuinely needs a human touch.
