Scaling support operations using generative AI on AWS with i-Tom Solutions

i-Tom Solutions provides a SaaS invoicing platform that is critical for over 170,000 small and medium businesses (SMBs) in Romania, handling essential automated accounting, regulatory compliance, and banking integration. As government-driven digital invoicing regulations accelerated platform adoption, the volume of customer support requests grew rapidly.

Anywhere from 200 and 600 written queries across multiple communication channels were arriving daily, from minor interventions or answers to technical audits and individual in-depth investigations, stretching the FGO team and causing delays that might affect customer satisfaction. i-Tom Solutions was eager to implement an intelligent virtual assistant that could automate routine responses and improve service efficiency.

After previous approaches had not yielded a production-ready platform, i-Tom Solutions turned to the proven and managed services of AWS, its cloud provider for over a decade. “The purpose of the solution is to have the agent by itself answer questions that arrive outside of business hours,” says Valeriu Filip, founder at i-Tom Solutions. “That was our first objective—to maintain responsiveness even when our team is offline.”

To address its scaling challenge, i-Tom Solutions worked alongside AWS solutions architects and AWS Partner Auvaria to design a POC. The goal was to create a retrieval-augmented generation agent that could answer customer questions by drawing from i-Tom Solutions’ centralized knowledge base.

The core of the architecture was built on Amazon Bedrock, which powers generative AI for more than 100,000 organizations worldwide. “Amazon Bedrock gave us the proven infrastructure and enterprise-grade security we needed to confidently take an AI agent from concept to production,” says Andi Aparaschivei, software architect at i-Tom Solutions. For document processing, the team used AWS Lambda, a service that empowers businesses to focus solely on code while it handles all infrastructure management. Semantic search was performed by Amazon OpenSearch Service, which simplifies AI-powered search, observability, and vector database operations.

i-Tom Solutions’ internal AI team collaborated with AWS experts during the POC phase to establish a reliable, replicable infrastructure. Using infrastructure-as-code templates, the company deployed the validated architecture in its production environment in just 1 week. The live assistant is currently being piloted with select customers while the company monitors its accuracy and refines its performance. “It proved to be the ideal scenario for us, working alongside AWS experts who guided us and made sure the project didn’t falter just before the finish line,” says Aparaschivei. That collaboration gave i-Tom Solutions the freedom to invent, using the comprehensive AWS AI portfolio to build a transformative customer experience.

i-Tom Solutions successfully brought its generative AI assistant to controlled production, delivering reliable automation at scale—an outcome that earlier attempts had not achieved. The assistant now manages hundreds of customer inquiries each day, reducing pressure on the FGO team and facilitating fast, consistent responses.

Building on that success, i-Tom Solutions plans to extend the capability across its platform, positioning AI-driven support as a defining feature of its service model. The company is already planning its next AI-powered enhancements, aiming to move beyond support into core product interactions. Now positioned to empower its customers further, i-Tom Solutions plans to introduce a module with which its customers can build their own AI agents and explore natural language interactions within its core invoicing platform. “Our intent is for customers to interact with their invoices using a natural language approach,” says Filip. “The next step is to interact with natural language instead of clicking and typing.”

By working with AWS, i-Tom Solutions has not only overcome an immediate operational challenge but also embedded generative AI as a core, scalable component of its future growth. “Working with AWS transformed our support scalability from an operational concern into a strategic platform feature,” says Filip.