Imagine your company carries more than 4,000 SKUs. Private buyers, dealers, and developers all call in, each with a different request. During peak season, call volume grows so fast that simply answering every call becomes a challenge in itself. Rather than expand its team of operators, VODALAND — which ran into exactly this problem — decided to hand its first line of calls over to an AI assistant. That’s how Ringostat’s AI Voice Agent joined the company. Here’s how it took over all inbound calls, and why it turned out to be nearly seven times cheaper than a contact center.
About The Company
VODALAND works with how water moves through cities, roads, industrial sites, and private properties. The company builds systems that collect, direct, treat, and store water — drainage channels and grates, manholes, wastewater treatment systems, tanks, and other engineering solutions for site development.
VODALAND has operated in this field for over 28 years and is part of an international manufacturing and engineering group present in seven countries. In Ukraine, the team includes more than 60 specialists, and the product range covers over 4,000 SKUs for different tasks and site scales.
The customer flow here is just as varied: private buyers, businesses, dealers, project organizations, and developers. Right from the first contact, the company needs to figure out who’s calling and which specialist should handle the request. VODALAND has been systematically digitizing its customer service — rolling out a CRM, automating processes, and looking for AI solutions that can respond to customers at any time of day.
Seasonality and Peak Load as the Main Challenge
Before launching the AI assistant, VODALAND already used Ringostat telephony integrated with its CRM. The e-commerce team handled initial call intake from the website, combining sales work with first-line duties.
The problem surfaced during seasonal peaks, when call volume grew quickly. The company considered three options:
- growing its own team,
- bringing in an external contact center,
- automating the first stage of communication with AI.
Hiring more operators would only solve the headcount problem. VODALAND needed to maintain consultation quality, response speed, and the ability to scale during peak season all at once. After weighing the costs, VODALAND chose Ringostat’s AI assistant, which the team named Eva.
Why VODALAND Chose Ringostat’s AI Voice Agent
VODALAND had worked with Ringostat for years and knew the platform and team well. The project lead in charge of solving the first-line problem during peak season had recommended Ringostat to his own clients even before joining VODALAND, and had helped implement telephony and call tracking at other companies. So when it came to choosing a partner for the AI assistant, Ringostat was the obvious choice.
Work on the concept began in early 2026, and the first working version of Eva launched in March.
What Eva Handles
The AI Voice Agent was given several specific tasks to automate the first line of customer communication:
- Qualifying the request. Eva asks clarifying questions, identifies the essence of the inquiry, and gathers the information needed to process it further.
- Initial consultation. The agent answers common questions and advises on product categories, drawing on the company’s knowledge base.
- AI call summary. After each call, Eva structures the information gathered and puts together a short summary of the conversation. That lets the team grasp the gist of a request without listening to the full recording every time.
- Round-the-clock availability. Eva handles calls 24/7, including outside business hours, on weekends, and on holidays. Customers can reach out whenever it suits them, and no request goes unanswered.
The AI Voice Agent mainly processes inbound calls from the VODALAND website.
According to Andrii Pyrih, Eva has effectively become an extra first-line operator — one who can back up the team during busy hours and keep service running after hours. In practice, plenty of customers call late in the evening and get a consultation even when the human team is unavailable.
Launching and Configuring the Campaign
VODALAND started by preparing a technical brief, and the initial setup took about a week. The most important phase began after launch: fine-tuning the dialogue logic and the agent’s rules of behavior for non-standard situations.
The team wrote the first version of the prompt on its own, using Ringostat’s examples, then reworked it significantly after test calls — each batch of conversations revealed which phrasing confused customers. It was a natural, iterative process: every round of calls surfaced new observations and pointed to rules that needed refining.
Today, a conversation with Eva follows this script:
- greeting,
- identifying the need,
- clarifying details,
- consultation,
- summary,
- wrapping up the call.
The team deliberately added a separate phase for summarizing the conversation.
Knowledge Base, AI Guardrails, and Tone of Voice
VODALAND already had plenty of material for customer-facing work — training documents, catalogs, technical documentation. But there was no ready-made knowledge base built specifically for a voice AI, so the team built one almost from scratch, adapting existing materials into a format suited for AI consultations.
Today, the knowledge base covers the company’s core product lines — drainage, manholes, landscaping, and other groups with their own product series. For VODALAND, it isn’t a one-off document but a living asset that the team keeps expanding as new scenarios come up.
A separate section of the prompt spells out not just what the AI should do, but what it shouldn’t: language models tend to fill in gaps when data is missing. So the team built in limits around unverified information and defined situations where the conversation needs to be handed off to a human.
The assistant’s voice turned out to matter just as much. During testing, the team tried several options but ultimately went back to the original one — within a few months, it had become a recognizable part of Eva’s identity.
Eva tells every caller upfront that she’s a virtual assistant. Some customers carry on the conversation just as naturally as they would with a human rep. Others, as soon as they hear the words “virtual assistant,” ask right away to be connected to a person. VODALAND has noticed a pattern: once a customer keeps talking and sees that the AI understands context and has a broad knowledge base, the conversation quickly starts to feel more natural.
Read also: How an AI Voice Agent Saves TI Over 190 Working Hours a Month
CRM Integration and the Personal Dashboard
CRM integration is a key part of the AI Voice Agent rollout. After each call, structured data appears in the system — data the team considers essential for handling the next stage quickly. That way, a sales rep immediately sees the context of the request and can pick up the conversation without asking the customer to repeat basic information.
Right now, data flows mostly in one direction — from Ringostat to the CRM. The next step is two-way exchange, so the AI can identify a customer by phone number during the call and see their order history.
The AI Voice Agent dashboard didn’t cause the team any trouble — Ringostat’s support team helped with setup from the start. VODALAND uses the built-in analytics sparingly, since the data is already analyzed in the CRM, but finds data export especially useful.
Results and Takeaways
Over the time it has used the AI Voice Agent, VODALAND has seen several key results:
- 7x lower costs than a contact center. Before launch, VODALAND collected quotes from external contact centers and compared operating costs. By the company’s estimate, the AI assistant turned out to be roughly seven times cheaper for a comparable volume of requests.
- 24/7 availability. At night, on weekends, and on holidays, Eva is the main point of contact; during the day, she backs up sales reps during peak hours.
- Zero missed calls. Even when every rep is busy, customers can still get through and get an answer.
- Faster work for reps. Instead of listening to the full recording, a rep gets a ready-made AI call summary.
- Consistent service quality. Service standards don’t depend on call volume or time of day.
There’s one more important point: scalability. VODALAND can expand first-line call capacity without proportionally growing headcount or going through lengthy hiring and training cycles for new operators. That matters most during peak season, when call volume spikes sharply.
VODALAND is also piloting another AI use case: analyzing sales reps’ phone conversations, currently tested in one department. The goal isn’t to review 100% of calls — some are short or purely transactional and don’t need deep analysis. Instead, the company wants to identify which call types, rep roles, and evaluation criteria deliver the most business value. For the first line of calls, plans include deeper CRM integration, conversation personalization based on customer history, and systematic transcript analysis.
Eva’s case shows that a successful AI rollout isn’t about plugging in an off-the-shelf service. It’s ongoing work: building out the knowledge base, refining the agent’s limits, and learning from real customer conversations.
