Case Studies | FinTech

From third-party APIs to enterprise-grade AI

Replacement of OpenAI with Bedrock

About

A FinTech business working with Cloud Combinator on AWS. The client is anonymised at their request.

Challenge

The engagement concentrated on three focus areas: removing dependency on external AI providers, preserving application performance under load, and doing so within a secure, cost-transparent AWS environment.

Dependency on third-party models

The application relied on OpenAI's embedding model and GPT-4 text model. Sending sensitive client and fund data to an external API sat uncomfortably with the enterprise-level security posture the client needed, and it left the business exposed to another provider's pricing, availability and data-handling decisions.

Performance and reliability under load

Invoking a language model for every client question, at scale, can trigger throttling and exception-handling problems that degrade the user experience. The solution needed a processing approach that kept responses reliable even as request volumes grew.

Security and cost transparency

As a financial services business, the client needed the model layer to run inside a secure AWS account with proper permissions, alongside a clear, predictable cost breakdown so the economics of the application could be understood and defended over time.

Solution

A client question arrives through API Gateway, which either accepts data directly from the client's application or integrates into the client's existing pipeline. The relevant content from the client's knowledge base is embedded and matched using an Amazon Bedrock embedding model, replacing the previous OpenAI embedding step, so the system can identify the most relevant source material for the question.

2

OpenAI models replaced by Amazon Bedrock (embedding and GPT-4 text)

$1,281

Projected monthly AWS run cost (MRR) for the Bedrock workload

50%

AWS cost contribution towards the engagement

By the numbers:

  • 2 - OpenAI models replaced by Amazon Bedrock (embedding and GPT-4 text)
  • $1,281 - Projected monthly AWS run cost (MRR) for the Bedrock workload
  • 50% - AWS cost contribution towards the engagement
Changes

Acceptance was based on signing off the Cloud Accelerator scope of delivery, confirming the technical competence of the client's internal tech lead, and demonstrating that the wider AWS environment is a suitable foundation for the client's further product development. The migration removed both external OpenAI dependencies and moved the AI layer into a secure, self-owned AWS platform.

  • Third-party dependency removed — Both the OpenAI embedding model and the GPT-4 text model were replaced with Amazon Bedrock, bringing the model layer inside the client's own AWS account.
  • Enterprise-grade security — The AI capability now runs within a secure environment with proper permissions, so sensitive client and fund data stays under the client's control.
  • Reliable at scale — Batch-style model invocation with SQS-based response handling reduces exception and throttling issues so the assistant stays responsive as usage grows.
  • Cost transparency — A detailed AWS cost breakdown gives the client predictable, defensible economics for the application, projected at roughly $15,372 per year.
  • Handover — A hands-on immersion day and best-practice foundation left the client's internal team able to operate and extend the platform independently after sign-off.

With the model layer now running natively on Amazon Bedrock inside their own secure account, the client has a foundation it fully controls for expanding the assistant to more of the knowledge base, experimenting with newer Bedrock models, and building further AI-driven products on top of the same proven pattern.

AWS Stack

Amazon API Gateway

For accepting requests from the client's application and integrating the solution into the client's existing pipeline.

Amazon Bedrock

For secure, enterprise-grade embedding and text generation, replacing the external OpenAI models.

AWS Lambda

For invoking Bedrock models with batch-style processing to keep responses reliable under load.

Amazon SQS

For buffering and delivering sets of model responses to reduce exception-handling performance issues.

YOU MIGHT LIKE

Related success stories

View all case studies

Case Studies | Insights

Utilising Language Recognition, Speed, and Enhanced Security to Make Social Media a Force for Good

  • Here, we take a detailed look at how the Cloud Combinator team collaborated with another cutting-edge AI service provider that provides intelligent systems to “make social media more social” for brands and users alike.
  • Arwen AI is a UK-based startup specialising in AI solutions to manage and enhance brands’ social media interactions. Founded in 2020 by Matt McGrory, Dr. David Cole, and Joel Bailey, Arwen. AI focuses on using AI to automatically detect and remove spam, toxic comments, and other unwanted content from social media platforms.
  • The team at Arwen have three core products. ‘Moderate’ is focused on identifying and removing toxic content from social media channels. ‘Engage’ helps brands identify and engage with meaningful conversations on social media, and ‘Customize’ allows brands to apply bespoke algorithms to their channels - creating an even more effective moderation and engagement.
Read more
CONTACT US

Ready to turn AI into impact?

We'll help you spot the highest-value opportunities, reduce risk around your first AI initiative, and define a clear path to results from day one.

Why talk to us:

Outcome-driven recommendations

AWS-recognised delivery expertise

Risk-aware AI adoption

Clear next step, not a sales pitch

Start with a focused 20-minute conversation about your goals — no pressure, no commitment.

We use cookies

This website uses cookies to enhance user experience and to analyze performance and traffic on our website.

See our Privacy Policy for details.