Редакція dev.ua AI Eng 2 February 2026, 10:40

On-screen prompts and voice robots: AI and other technologies that insure businesses against human errors

A call to a clinic: the operator cannot immediately make an appointment because the Internet is slow due to shelling, and promises to call back. They say goodbye, and the patient waits for a call. The operator has to record the lead in the CRM, look at the doctor’s schedule, and offer the patient an appointment. But he switches to new requests and forgets about the promise. In conditions of such multitasking and stress, even an attentive employee can make a mistake.

For businesses, these are constant risks: lost customers, broken agreements, reputational damage. It’s not just about medicine — similar scenarios are repeated in other businesses. However, often it’s not the fault of the employee, but the result of «manual» processes.

dev.ua, together with IPTel, a developer of systems for communications automation, tells what technologies help «insure» businesses against human errors and how it works in practice.

How to stop losing customer data

The story of a lost patient is a typical situation for a service business. Last year, the European dental network Dynasty Stomatology approached IPTel with a problem: with the increase in the number of clients, it became difficult to maintain the quality of communication at a high level — managers did not have time to process calls. And since telephony did not support integration with CRM, all contacts had to be entered manually. As a result, data was lost, and the processing of applications was slowed down.

The first step towards automation was the implementation of IVR — an interactive voice menu, which takes on the task of receiving and coordinating client requests.

Previously, the manager had to clarify the request each time and manually transfer the call to the desired department. Now the system does this automatically: as soon as the patient dials the clinic number, a voice menu is activated, where he selects the type of request that corresponds to his request — and immediately receives the necessary information.

«IVRs allow you to build multi-level conversation scenarios and perform some routine requests without human intervention,» says IPTel CEO Ivan Vinogradov. «The system can work via tone dialing, when the client presses a key on the phone, or voice recognition: the client says, 'I want to make an appointment with a therapist' — and it understands the request.»

Since the IVR integrates with the CRM system, it has solved a key need for the clinic to organize customer data, information about appointments and calls. If the patient calls again, the system will automatically «recognize» him by phone number, pull up a card with the history of visits and can even greet him by name.

For the Dynasty Stomatology dental network, which has branches in Poland and the Czech Republic, IPTel engineers created a multi-level voice menu that offers to choose a branch and type of request. Depending on the choice, it connects with a manager. Now clients always receive an answer and get to the right specialist, and managers have fewer routine tasks.

When one phrase can influence a decision

One of the Ukrainian online electronics stores turned to IPTel with a problem: despite active marketing efforts, sales were not growing. Management had reason to suspect the human factor. When the company connected a speech analytics tool to analyze the quality of conversations with customers, the suspicions were confirmed. During telephone consultations, managers did not voice the key characteristics of the products — precisely those that, according to the company, significantly increased the likelihood of purchase.

«To control the quality of conversations, you need to listen to them. These are hours of recordings that managers often don’t have time for physically,» explains IPTel CEO Ivan Vinogradov. «And if you check selectively, a call where the manager made a mistake may not be included in the sample. The problem will go unnoticed until the client writes a complaint about being rude to him.»

AI-based speech analytics solves this problem. The manager sets criteria for what the manager should say during the conversation and assigns each a score. For example, describing all the characteristics of the product — 100 points, suggesting a related product — 30 points. The tool transcribes conversations and analyzes them according to these criteria in a matter of minutes. If the manager missed something — no points are awarded. The result of the analysis is a report with statistics for each employee.

The tool works on large language models (LLM). The choice depends on the task: where the conversation recordings may contain sensitive data, the IPTel team uses local models that can be deployed on the client’s infrastructure, such as DeepSeek. For other tasks, cloud solutions based on OpenAI or Gemini are used.

«The key to speech analytics is stability, accuracy, and control of hallucinations so that the AI doesn’t invent things that weren’t in the conversation,» says the CEO of IPTel. «We managed to achieve recognition accuracy of 95%.»

AI assistant to speed up contact centers

Previous tools automate routine. In fact, it is insurance at the first stage of communication and «work on mistakes» after. The next step is to help the person in the moment.

«During a conversation with a client, the operator is left alone with his request. Of course, he has a script, but a person can ask a question that is not there,» says the CEO of IPTel. «So we thought: what if we gave the manager an assistant who would prompt the correct answers right during the conversation?»

IPTel AI assistant gives tips to the operator

IPTel’s AI assistant can be connected to the company’s knowledge base — internal documents, regulations, reference materials. During a conversation, a manager can ask a question, as in ChatGPT, and instantly receive an answer based on corporate data. He does not need to keep dozens of documents in his head or search for information in different files at the same time. The effectiveness of an AI assistant depends on the quality of the data. If the company’s information is unstructured, the system will not be able to give a correct answer.

In one of the private clinics in Odessa, the IPTel team implemented an AI assistant for contact center operators. It does not replace a person, but works alongside them: it analyzes the request and searches for answers in the internal document database. As a result, the average application processing time was reduced by approximately 25%.

At the same time, the hospital did not go the route of reducing staff. On the contrary, highly qualified employees were transferred to more complex areas, in particular, work with insurance companies, and the routine part of consultations was left to less experienced operators, enhanced by AI.

«Communications automation is not about replacing people, but about strengthening businesses where the human factor begins to affect the quality of service,» concludes IPTel CEO Ivan Vinogradov. «When routine processes are taken over by technology, companies gain stability and the ability to scale. And people can focus on what really requires their participation.»