By automating translation, categorization, and routing of ~250 weekly incoming support messages, an international IT company reduced the actual effort from 1,000 hours per month (~1 hour per request) to 4 hours (one specialist once a week). This eliminated "stuck" requests, accelerated response time from hours to minutes, identified 11 potential partners, and saved 856,400 RUB monthly.
Result:
- Savings: 856,400 RUB monthly
- Acceleration: 996 hours of work per month
Client
An international software development company for freight transportation monitoring, operating in 9 countries.
Objectives
- Unify the processing of support inquiries across multiple languages (English, Turkish, Arabic).
- Automatically distribute requests by topic and department across the company.
- Identify potential partnership inquiries among customer requests.
- Optimize the workload on sales department employees and increase request processing speed.
Initial Situation
- About 250 messages per week arrived in the support chat in different languages.
- Requests required translation, topic-based routing, and distribution to departments.
- Request processing was handled by sales department employees in addition to their primary duties.
- Due to human error, some valuable partnership requests were lost.
Deployed AI Solution
- Development of a neural-network-based algorithm for automatic translation of requests into a single language.
- Creation of prompts for categorizing requests by defined criteria and topics.
- Isolation of atypical requests for additional analysis and category expansion.
- Retrospective analysis of past inquiries to identify patterns in non-standard requests.
- Development of a keyword-based request labeling algorithm for CRM routing.
- Preparation of technical specifications for integrating the AI solution with the company's CRM via API.
Results
- Automated processing and routing of 250+ weekly support requests.
- Identified 11 potential partnership requests from companies across different countries.
- Reduced workload on sales department employees; a dedicated specialist was assigned to request processing (1 hour per week).
- Accelerated response to typical customer requests through automatic site link delivery.
- Improved customer satisfaction due to fast and accurate responses to inquiries.
Conclusions
Deploying AI for support request processing enabled the international IT company to unify cross-language communications and improve the speed and quality of customer inquiry responses. Automatic categorization and routing reduced employee workload, while identifying potential partnerships opened new business development opportunities. The AI solution scales through integration with the company's CRM system.