In his opinion, these areas of application of language

[tanjimajuha20]

a Shchirov, Product Director, Head of AI Development and Support Products at Raiffeisenbank, described three scenarios for using generative artificial intelligence (AI) at the DataDay forum on July 4, which are effective from a commercial point of view: first, generating datasets using generative models for chatbots. For this, according to Ilya Shchirov, open source can be used. The risks will be minimal, and the tool is easily monetized.

"The second application is spain whatsapp resource the integration of large language models (LLM) into CRM systems. This is necessary to summarize not individual elements in text, but the entire customer journey. CRM provides omnichannel, it displays all points of contact between the customer and the service. But a specialist still needs time to read the information and understand. There is omnichannel, but there is no sense. A generative model should quickly describe the customer journey. There will be a huge breakthrough, and we are working on this, when the generative model manages to briefly outline the customer journey and help the operator. This is a difficult task, but not so expensive. The main problem lies in the data," noted Ilya Shchirov.

The third case, according to him, is the use of generative models in knowledge bases, which are available in almost all large banks and help specialists in the support department. "Here, the model can give not just a list of search results from the knowledge base, but a specific answer. This should be an auxiliary tool for those specialists who work with people," added Ilya Shchirov.

models are low-risk and easily calculated in terms of costs and benefits.

Valentin Malykh, Associate Professor at the Higher School of Digital Culture at ITMO University, believes that if a company does not have IT competence, then perhaps it is not worth creating a Data Science department from scratch.

"The Data Science department is a large project. It is necessary to find a manager who is ready to take on this, as well as employees who are capable of doing the work. This should be a continuous process, which is quite difficult. One of our clients, a large mining company, has a Data Science department that solves certain problems, but it is small. The company employs about 10 thousand people, and the Data Science department has about 10,000, and they cannot solve all the problems required for such a large company, so some competencies need to be outsourced," says Valentin Malykh