Case study: AI-powered secure financial advisory platform

Customer Overview

Customer	BankowyNegocjator
Industry	Financial Services / FinTech
Location	Poland
Type	SMB

Executive Summary

bankowynegocjator.pl, a Polish financial advisory company, wanted to leverage generative AI to enhance its client advisory services – providing faster, more comprehensive financial analysis and recommendations. We as Tehma Cloud deployed an AI chat platform powered by Amazon Bedrock on a highly available, cost-optimized AWS infrastructure, enabling the company to offer AI-assisted financial guidance while maintaining full control over data and model selection.

Challenge

Financial advisors spent significant time researching and preparing personalized recommendations for each client consultation.
The company needed access to multiple AI models to compare outputs and select the best response for different financial scenarios.
Data privacy was paramount – client financial information could not be sent to third-party AI providers without strict controls.
Budget constraints required an infrastructure solution that minimized fixed costs while maintaining high availability.

Solution

We deployed an open source AI chat platform integrated with Amazon Bedrock through a compatibility gateway, enabling access to multiple foundation models (Claude, Titan) through a unified interface. The platform runs on Amazon ECS with a mixed instances policy spanning multiple ARM64 Graviton instance types (c6g, m6g, r6g), using a price-capacity-optimized strategy for maximum cost efficiency.

Workflow automation via n8n enables the advisory team to build custom AI-powered workflows – such as automated document analysis, client briefing generation, and regulatory compliance checks – without writing code.

The infrastructure includes Amazon RDS (PostgreSQL) for persistent data, Amazon EFS for file storage, and a vector database for semantic search across financial documents. All critical components run in private subnets with no public internet exposure. The platform communicates with Amazon Bedrock within AWS VPC Endpoints (Privatelinks) so that even the AI traffic is hidden within private channel. Data stored by the platform lays in regional S3 buckets preserving data residency and GDPR compliance.

Amazon Bedrock gives us access to the latest AI models without managing any self hosted infrastructure. Our advisors can now prepare client recommendations in minutes instead of hours.

Dr Marcin Korulczyk – Founder of Bankowy Negocjator

Results

5x faster client preparation – AI-assisted research and recommendation generation reduces preparation time from hours to minutes.
Multi-model flexibility – access to multiple foundation models through Amazon Bedrock enables selecting the best model for each use case.
Low cost infrastructure – mixed Spot instances and ARM64 architecture minimize fixed costs.
Full data control – all processing occurs within the customer’s AWS account and AWS private network with no data leaving the environment or exposed to public Internet.
No-code automation – workflow automation enables non-technical staff to build and modify AI-powered processes.

AWS Services Used