ECONOMIC EFFICIENCY AND TECHNOLOGICAL CHALLENGES OF FULL SUPPORT AUTOMATION: FROM EXPERT INTERVIEWS TO AUTONOMOUS AI AGENTS
Received: 2026-07-17 16:27:17
Published: 2026-04-18
Abstract
The article is devoted to the study of the economic efficiency and technological challenges of fully automating customer support using autonomous AI agents based on large language models (LLMs). In the context of the digitalization of the financial technology sector, traditional customer service models face increasing operational costs, limited scalability, and dependence on the human factor. The paper examines a practical case of implementing an autonomous support system in a fintech company based on a hybrid architecture that includes a verified knowledge base, multi-level orchestration of language models, and escalation mechanisms for complex requests to human operators. Particular attention is paid to the methodology of transforming organizational expert knowledge into a structured AI-compatible knowledge base through expert interviews and subsequent data processing using the Anthropic and OpenAI model stack as well as cloud-based AI infrastructure. The empirical section presents the results of the industrial implementation of the system at Analytics Services, which made it possible to automate up to 90% of customer requests, reduce operational support costs by 20 times, and ensure industrial operation of the system without critically significant errors. Based on the obtained results, the article concludes that with proper architectural implementation, autonomous AI agents can function as полноценная digital units within a company’s customer service structure.
Keywords
List of references
-
Russell S., Norvig P. Artificial Intelligence: A Modern Approach. Pearson, 2021.
-
Goodfellow I., Bengio Y., Courville A. Deep Learning. MIT Press, 2016.
-
Vaswani A. et al. Attention Is All You Need. NeurIPS, 2017.
-
Brown T. et al. Language Models are Few-Shot Learners. NeurIPS, 2020.
-
OpenAI. GPT-4 Technical Report. 2023.
-
Zubayda Jumayeva, Nurbek Khushvaktov, Rustam Nizomov. BIO Web Conf., 173 (2025) 01030. DOI:https://doi.org/10.1051/bioconf/202517301030
-
Anthropic. Constitutional AI: Harmlessness from AI Feedback. 2022.
-
Lewis P. et al. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.
-
Shneiderman B. Human-Centered AI. Oxford University Press, 2022.
-
Davenport T., Ronanki R. Artificial Intelligence for the Real World. Harvard Business Review, 2018.
-
Brynjolfsson E., McAfee A. The Second Machine Age. 2014.
-
Kaplan A., Haenlein M. Business Horizons, 2019.
-
Gartner Research. Autonomous AI Agents in Enterprise Operations. 2024.
About the Authors
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
How to Cite
