Case Studies | Retail

Forecasting demand and taming the inbox

Forecasting and Summarisation Backend

About

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

Challenge

The challenge had four focus areas.

Inventory guesswork

Deciding when to buy, what to buy and in what quantity, so that a venue holds an acceptable but not excessive amount of stock, is a hard judgement to make by hand. Get it wrong and you either run out or tie up cash in stock that spoils, so the system needed to forecast demand and translate it into buying recommendations.

Data scattered across sources

The inputs live in different places, from sales figures and stock counts to written material. The backend had to ingest a diverse range of datasets and bring them together so predictions reflect the full picture rather than one silo.

Information overload

Beyond numbers, the team faces long-form text in emails, presentations and documents. The solution needed to summarise these accurately, and where several sources are involved, aggregate them into a single final result rather than a stack of separate answers.

Cost-efficient integration

Whatever was built had to sit alongside the client's existing frontend and AWS stack, return results in a timely fashion, and keep running costs down, which shaped an architecture of small, single-purpose functions rather than one heavy service.

Solution

An end-user interacts with the client's frontend, and either on page access or through an UI element triggers the relevant AWS Lambda function through its Function URL. Each Lambda is a handler for a specific feature or metric, kept deliberately lightweight to reduce cost per invocation, with an API Gateway routing requests to the right function based on the REST API path.

10-day

Engineering build, deployed as CloudFormation infrastructure-as-code

Serverless

A lightweight Lambda per metric to keep per-invocation cost low

RAG

Amazon Bedrock retrieve-and-generate over a S3 Vectors knowledge base

By the numbers:

  • 10-day - Engineering build, deployed as CloudFormation infrastructure-as-code
  • Serverless - A lightweight Lambda per metric to keep per-invocation cost low
  • RAG - Amazon Bedrock retrieve-and-generate over a S3 Vectors knowledge base
Changes

The engagement delivers a working forecasting and summarisation backend that produces predictions in line with current trends and accurate, context-aware summaries of long-form content, returned to the client's frontend through a clean API. Acceptance is on demonstrating that the solution forecasts reliably, summarises accurately across multiple sources, and responds in a timely fashion.

  • Demand forecasting — The backend ingests sales and stock data to estimate when and what to buy, and in what quantity, to keep inventory acceptable but not excessive.
  • Document summarisation — Long-form emails, presentations and reports are summarised accurately, with multiple sources aggregated into a single final result.
  • Modular by design — A separate lightweight Lambda per feature, routed by API Gateway, keeps each function cheap to run and makes new metrics easy to add.
  • Grounded answers — Amazon Bedrock retrieve-and-generate over an Amazon S3 Vectors knowledge base keeps responses accurate to both the metrics and the context.
  • Handover — AWS CloudFormation, a demonstration and Loom walkthroughs leave the client able to deploy, operate and extend the backend independently.

With a modular, serverless forecasting and summarisation backend now sitting inside its AWS stack, the client has a foundation it can keep extending, one new Lambda at a time, as more metrics and features are added for the venues it serves. The same knowledge-base pattern is ready to power whatever data-driven capability the client wants to offer next.

AWS Stack

Amazon Bedrock

For forecasting and summarisation via a retrieve-and-generate pattern.

Amazon S3 Vectors

For the vector store underpinning the knowledge base.

AWS Lambda

For lightweight, single-purpose functions behind Function URLs.

Amazon API Gateway

For routing REST requests to the correct function.

AWS Identity

And Access Management for least-privilege roles per function.

AWS CloudFormation

For automated, repeatable deployment of the stack.

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.