MTIBU (Motor Transport Insurance Bureau of Ukraine)
MTIBU (Motor Transport Insurance Bureau of Ukraine) handles a high volume of inquiries from insurers, claimants, and partners regarding MTPL (OSTSV) insurance conditions, regulatory requirements, and claim procedures. Agents had to manually search normative documents for each request, increasing contact center workload, slowing response times, and reducing service quality.
There was no automated way to detect and escalate time-sensitive critical cases – such as accidents with injured parties – in real time, and no way for users to get accurate, source-grounded answers outside business hours. As inquiry volume grew, MTIBU’s contact center faced rising operational cost with no scalable path forward.
Tech-5 designed and deployed a fully serverless AI chatbot for MTIBU, using Amazon Bedrock Knowledge Bases and Amazon OpenSearch Serverless (with Titan Text Embeddings v2) to ground every response in MTIBU’s curated regulatory documents, with mandatory citations to the source material.
Incoming queries are classified by intent and routed across Amazon Bedrock foundation models – Amazon Nova 2 Lite for lightweight conversational intents, and Anthropic’s Claude Haiku 4.5 for substantive regulatory queries – balancing response quality, latency, and cost. Out-of-scope queries return a clear fallback message directing users to the contact center, and critical inquiries are automatically detected and escalated with relevant operator contact details. MTIBU administrators manage and update the knowledge base directly through a dedicated admin dashboard, without ongoing developer involvement.
MTPL regulatory queries are now resolved automatically, around the clock, with responses grounded in official MTIBU documentation and always accompanied by source references. Critical, time-sensitive inquiries are detected and escalated automatically, ensuring timely human follow-up where it matters most. With self-service knowledge base management in place, MTIBU can keep the assistant’s regulatory content current without depending on ongoing developer support – and the underlying serverless AWS architecture is designed to absorb growing inquiry volume without proportional increases in contact-center staffing.