AI call center: how Uzbekistan’s top telecom built a high-performance CC

AI call centers are redefining customer service across industries — and Sarkor Telecom's journey shows precisely how this transformation works in practice. Sarkor Telecom is one of the leading telecommunications providers in Uzbekistan, operating since 1996 and ranking among the top three market players. The company holds approximately 15% of the national market share and serves both residential and corporate customers across the nation.

Sarkor's portfolio includes high-speed GPON and FTTx internet, IPTV, IP telephony, cloud services, hosting, and video surveillance. Thanks to its extensive infrastructure and reliable service, the brand has become one of the most recognizable names in the country's telecom market.

However, as the business expanded and introduced more complex products, maintaining flawless customer service became an increasingly challenging task. Traditional contact center operations could no longer keep up with the workload and rising customer expectations. For large companies with well-established processes, any major technological shift carries significant risks: it alters familiar workflows, necessitates staff retraining, and demands seamless integration of new tools into existing systems.

Sarkor, however, made a bold move — to fully embrace customer service automation by launching an AI Call Center project. For this, the company turned to Apifonica, a trusted AI solutions provider specializing in customer communication automation. At the core of the project were intelligent AI agents capable of handling multilingual inquiries, processing routine requests, and integrating seamlessly into the existing customer service ecosystem, all while maintaining the customer experience.

In a conversation with Inessa Malkova, Director of the Retail Customer Service Department, we uncovered valuable insights that could benefit our future clients — and we're sharing them here.

What challenges did Sarkor's contact center face

For a large telecom provider with a customer base in the hundreds of thousands, keeping the contact center running smoothly is mission-critical. In reality, however, service representatives often worked under constant pressure. Under normal conditions, the team can handle up to 1,200 calls per day; however, during outages or technical failures, this volume can spike by a factor of 17. The phone lines would become overloaded, wait times would increase, and some customers would be unable to reach a representative. A large share of this load came from repetitive, low-complexity questions, which consumed significant time and distracted agents from urgent, high-priority calls.

The contact center employs 40 people (not all of whom handle inbound calls) and previously operated in three shifts. During peak loads, the company had to mobilize additional resources to maintain 24/7 operations — yet even this didn't guarantee a prompt response for every customer. The average wait time exceeded two minutes, and during peak moments, it was significantly longer. Additionally, there was a noticeable level of subjectivity in how some representatives handled customer support, which affected consistency.

All of this made the job exhausting for service reps and the customer experience unpredictable — a sensitive issue for a primary telecom provider, as it shaped public perception of the company and fueled customer churn.

During the project launch, automation with voice bots uncovered an issue that had previously gone unnoticed: only 10% of subscribers in the CRM had accurate, up-to-date contact information. The efficiency of internal processes was directly affected, as many cases could not execute service scenarios correctly due to missing or inaccurate data.

Building an AI call center that delivers results: from concept to implementation

Sarkor designed its AI call center to handle peak loads without compromising service quality. Before the introduction of AI, all incoming calls went into a single queue, and customer service agents responded to them in sequence. With AI call center solutions in place, the system automatically routes traffic during peak loads: the platform sends calls related to outages, large-scale technical issues, or other repetitive requests into a separate flow, where an AI customer service agent handles them using a predefined algorithm.

In these AI workflows, the system categorizes requests, generates follow-up questions, and provides answers—all powered by "intents" that dynamically shape responses based on the context of the request. This setup enables customers to access the information they need without waiting for a live representative, allowing specialists to focus on more complex cases.

As a result, they reduced the average wait time from more than two minutes to just 40-60 seconds, and they improved CRM data accuracy fourfold in just a couple of months. In the past, agents had big red "action" buttons to update CRM records. Still, attention dispersion caused them to neglect these processes, leading to outcomes that heavily depended on human factors. Now, improved CRM hygiene has driven sales conversion rates from 30% to 65%, effectively doubling them.

The introduction of AI call center agents delivered a tangible impact for both customers and internal contact center operations. Customer satisfaction rose from 78% to 91%, reflecting reduced wait times, more accurate answers, and fewer errors. At the same time, agents now spend 30% less time on routine tasks, allowing them to dedicate more effort to complex and high-priority issues. This means automation has brought a dual benefit: improving the customer experience while boosting the company's operational efficiency.

Was the implementation difficult?

Sarkor is a large organization, so preparing for the launch of the AI Call Center took time. Planning alone took about six months, including process alignment, defining call scenarios, and identifying integration points. The actual technical implementation, however, was much faster — visually, setting up workflows is not a highly complex task.

The process began with Apifonica's specialists designing the initial set of call-handling workflows and launching the technical implementation. After that, Sarkor's staff took over this part of the AI call center implementation, refining and adapting the workflows to better address real customer needs.

The contact center team has established workflows to offer multilingual support in Uzbek, Russian, and English. AI-driven call center technology now operates seamlessly in 95% of cases. Additionally, advanced conversational AI features implemented by Apifonica enable seamless handling of mixed-language conversations with minimal errors. These features allow the call center to adapt to the customer's language in real-time.

After implementing AI, was everyone laid off?

Spoiler: No one was laid off, which aligns with both Apifonica's and Sarkor's approach.

Stories about mass layoffs in contact centers went viral online at the dawn of generative AI, when some companies dismissed their entire support teams. We don't advocate for that level of extreme "AI-first" fanaticism. However, we recognize that optimization is a natural part of business, and staff reductions can occur for many reasons — including higher levels of automation.

The fundamental goal of automation projects is not to cut staff, but to optimize processes. In large companies, internal mobility and in-house recruitment are standard practices. Department heads often look to the contact center first when hiring, as it's where they'll find the most engaged employees who understand internal processes inside and out. On the open market, such people are either hard to find or expensive and time-consuming to recruit.

As a result, contact center staff often transition to other departments where their experience and skills remain in high demand. For those who stay in the contact center, AI adoption has become an opportunity for professional growth. By mastering AI tools and automation systems, they significantly increase their market value — a specialist who knows how to work effectively with AI agents will easily find a role, even in the most competitive industries, such as telecom, banking, and IT.

Beyond customer service: AI agent for high-volume hiring

Sarkor initiated a series of experiments with AI agents, which included developing a distinct workflow focused on automating bulk hiring. This workflow is specifically applicable to "field" agency channels, where workers, starting from the age of 16, distribute flyers or work in various departments under civil law contracts.

The challenge with these channels is that they require constant recruitment, but the HR department operates on a standard Monday-to-Friday, 9:00 a.m.–6:00 p.m. schedule, making it difficult to connect with candidates in real-time.

The AI workflow solves this problem: outside of HR's working hours, it automatically invites candidates to interviews and books them directly into a shared Google Sheet. This means that by the start of the next business day, HR already has a ready-made list of confirmed candidates — and can immediately proceed with the hiring process without any extra follow-up work.

Conclusion: AI agents as a strategic business transformation

Sarkor's experience with the AI-based call center shows that implementing AI agents is not just a one-time technical upgrade — it's a strategic move that impacts the entire business. What began as an initiative to relieve customer service representatives during peak hours and provide customers with faster, more accurate answers ultimately led to large-scale changes: improved CRM data quality, reduced wait times, higher sales conversion rates, and even the automation of processes not directly related to customer service — such as high-volume recruitment.

A key takeaway: automation did not become a threat to employees. On the contrary, AI-enabled workforce redeployment within the company, upskilling of remaining specialists, and a new level of business agility.

The Sarkor story is a clear example of how a primary telecom provider, operating in a conservative and competitive market, can take a calculated risk, entrust key processes to AI agents, and ultimately reach a new level of efficiency.

Watch our full interview with Inessa:

FAQ:

Will AI replace call center agents?

No. In Sarkor's case, no one was laid off after the implementation of the AI call center. The goal of automation was not staff reduction but process optimization

How does AI work in a call center?

At Sarkor, AI automatically routes calls during peak loads. Requests related to outages, technical issues, or other repetitive questions are directed into a separate flow, where they are handled by an AI agent following a predefined algorithm. The AI categorizes requests, asks follow-up questions, and provides answers based on "intents" that adapt to the context of the request.

How to use AI in a call center?

AI can handle repetitive requests, reduce customer wait times, and allow human agents to focus on complex cases. Sarkor also uses AI to enhance CRM data quality and automate bulk recruitment campaigns.

How to optimize international call center workflows with AI-powered agent assist?

Sarkor configured AI workflows for multilingual support in Uzbek, Russian, and English, achieving a 95% accuracy rate. The AI can handle mixed-language conversations (where the customer switches languages mid-call) almost flawlessly, making it an effective tool for optimizing international and multilingual call centers.