At peak load, half their clients hung up unanswered. Now, 4 in 10 are resolved before a human picks up.
For telecom providers, client support demand can spike dramatically during service disruptions. And even in day-to-day operations, high volumes of routine enquiries put constant pressure on contact centres.
For Sarkor Telecom, a major incident could trigger 17,000 calls in a single day, paralysing the contact centre and leaving 50% of clients in the dark.
By implementing a functional Agentic AI, Sarkor Telecom filtered calls and automated 40,000 minutes of human labour per month and uncovered critical business insights that were previously invisible to management.
The Company
Sarkor Telecom, founded in 1996, is the largest alternative internet service provider in Uzbekistan, serving around 100,000 B2C and 5,000 B2B clients. The company holds approximately 15% of the national market share and serves both residential and corporate clients across the nation. It has consistently driven innovation in the telecom sector, introducing services such as hosting, IPTV, IP video surveillance and cloud PBX. Sarkor Telecom was also the first in Uzbekistan to deploy an international fibre-optic channel and continues to expand its own extensive fibre network in Tashkent while pioneering GPON technology to deliver high-speed, stable internet with reduced dependency on electrical infrastructure. Thanks to its extensive infrastructure and reliable service, the brand has become one of the most recognisable names in the country’s telecom market.
The Challenge: Scaling client support without delays
Sarkor Telecom’s contact centre employs 40 staff members who operate in three shifts, not all dedicated to inbound call handling. As client demand increased, they needed a way to scale first-line support: under normal conditions, the team can handle up to 1,200 calls per day, but during network outages inbound volume could spike from around 1,100 to 17,000 calls within hours, leaving up to half of clients without a resolution. To manage demand, the company had to scale resources to handle the enquiry volume. However, this still did not ensure consistent response-time performance for all clients.
- Multilingual support in Uzbek, Russian, and English
- Roughly 40% of technical calls were “Level 0” resets, resolvable without a human, but still landing on specialist technicians.
- 2 minutes per call were lost to manual identity verification because CRM records were routinely out of date.
- Management had no visibility into the real causes of client frustration. Complaints were logged but never analysed at scale.
The Solution: An AI Agent that takes actions
Instead of relying on a menu-driven phone system, Sarkor Telecom deployed a functional AI Agent integrated directly into their billing system, CRM and network infrastructure.
- Instant caller recognition. The Agent identifies the client, checks their account balance and detects any active outage at their address before the conversation begins.
- Real-time resolution. The Agent doesn’t just advise; it acts. It can restart a client’s connection or log a support request on the spot, without ever passing the call to a human.
- Bilingual, natural-language handling with mid-sentence language switching and tolerance for informal speech patterns, built for regional nuances.
Unexpected outcome: What AI Revealed About the Real Causes of client Complaints
Sarkor Telecom initially believed its main challenge was the heavy load in its internal contact centre, specifically the surge in inbound traffic related to technical reports. However, the analysis of thousands of AI-handled call transcripts revealed a different picture. About one third of calls were related to routine modem restarts and another third concerned Wi-Fi connectivity and setup. In other words, the majority of inbound requests were routine issues that could be identified and resolved by the AI Voice Agent using standard scenarios without escalating to human support.
We assumed the complaints were purely technical. The data told us otherwise, and pointed us toward a service culture improvement we never would have found manually.
Inessa Malkova, B2C Division Director
More importantly, the analysis exposed a strategic blind spot in field technician service and communication quality issues. This shifted the focus from fixing infrastructure alone to improving the client experience and service culture.
4.1. CRM Debt and Operational Blind Spots Exposed by AI Analysis
The CRM system of Sarkor Telecom emerged as a significant operational constraint. Up to 2 minutes per call were lost to manual identity verification due to outdated and inconsistent client data, particularly phone numbers that had changed over time. This prevented automated caller recognition and forced agents to manually check contract details, verify session status and guide users through basic troubleshooting.
To resolve this, the AI Agent was integrated with the CRM, which helped to instantly recognise callers and check for location-specific network incidents. In outage scenarios, the system can inform clients of the issue and expected resolution time without involving an operator. Additionally, all AI interactions are automatically logged and classified, improving data quality, visibility and ongoing service optimisation.
Results: Measurable impact from month one
| Figure | Metric |
|---|---|
| Over 100,000 | Call minutes handled monthly by AI Voice Agent |
| 65% | First-line support calls fully automated |
| 95% | Cases handled end-to-end by AI Voice Agent |
| 0 | No wait time, no queues. Instant response at any volume |
| ~85% | Successful query resolution, post-CRM cleanup |
| 15-20% | Reduction in technician dispatch |
| 78% to 91% | Increased client satisfaction (CSAT) |
Further development
“We’ve transitioned Voice Agent management fully in-house and now operate as a self-sufficient team. All dialogue designs, scenarios and call flows are developed internally, with no dependence on external support to meet performance targets. We are also preparing to launch outbound campaigns, but entirely by our team using a no-code interface”
Inessa Malkova, B2C Division Director at Sarkor Telecom
The distinguishing feature of the interface is the ability to automate, structure and control conversations. The platform ensures that dialogues follow a predictable, business-defined path rather than an arbitrary AI-generated flow, which gives full control over dialogue. A goal management system allows tracking business-level metrics per conversation block, which many competitors lack.
Analytics reports and conversation transcripts make performance immediately visible, which simplifies AI agent management and removes the need for manual call monitoring.
Why it works: Three principles behind the results
Based on the Sarkor Telecom implementation, our AI Agent offers three core strategic advantages:
1. Scaling without re-engineering — The AI Agent connects to your existing billing and CRM infrastructure. Routine tasks are automated without requiring a system overhaul. 2. Consistent performance under pressure — Crisis spikes are where reputations are won or lost. The AI handles volume surges so your human team stays focused on the clients who need them most. 3. Automation as an investment, not a cost cut — Sarkor Telecom reinvested their automation savings into higher salaries and specialist training, turning their support centre into a revenue-contributing competency hub.
Ready to reclaim 40% of your agents’ time?
By the time a human agent would have finished verifying a contract number, our AI Agent has already diagnosed the connection, performed a remote reboot and confirmed the client is back online.
Last updated: May 2026
