Welcome to Silicon Overdrive!

Linkedin Youtube

Welcome to Silicon Overdrive!

Datafy: Enabling Intelligent Shelf Audits with AWS

About the Customer

Datafy develops retail execution and analytics solutions that give brands real-time visibility into product performance and availability at the point of sale. In retail environments, accurate shelf data directly influences revenue and brand perception, making reliable, timely audit information essential to Datafy’s value proposition. 

To improve the reliability and scalability of in-store audits, Datafy sought to replace manual inspection processes with a digitised, automated shelf analysis solution.

The Challenge

Datafy’s audit operation was built entirely on WhatsApp and Excel. Field representatives completed store visits and submitted results informally via WhatsApp, while a team of five or more administrators manually reconciled those messages, assigned follow-up tasks, and maintained shared spreadsheets.  

This weekly cycle meant that data was never available in real time, and the volume of messages and photographs generated persistent miscommunication between field and office teams. 

This approach was not scalable. Datafy required a purpose-built mobile application that would enable field representatives to capture shelf images during store visits and automatically assess product availability and compliance. 

Specifically, the solution needed to allow representatives to: 

  • Visit stores and capture shelf images using a mobile device
  • Upload images reliably in environments with limited or intermittent connectivity
  • Associate each image accurately with the correct store, visit, and product category

From an analysis perspective, the system needed to:

  • Process captured images to determine whether products were correctly stocked against task data derived from stock and sales records
  • Detect missing, misplaced, or incomplete product placements, including empty shelf space and gaps
  • Return a clear, actionable compliance result for each shelf: Fixed (fully stocked and compliant) or Not Fixed (requires attention)

Additional non-functional requirements included minimal disruption to in-store workflows, consistent and repeatable analysis results, and secure handling of all image data and associated metadata.

The Solution

Silicon Overdrive designed and delivered a mobile-first shelf audit platform combining offline data capture with AI-powered image analysis and centralised reporting.

Mobile Application and Web Portal

A cross-platform mobile application and web portal were developed using Flutter and Dart, enabling a consistent user experience across device types from a single codebase. The mobile application guides field representatives through the image capture process, enforcing consistent image quality while automatically associating each capture with the relevant store, visit, and product category.

Offline-First Architecture

Recognising that in-store connectivity is often unreliable, the solution was built on an offline-first architecture with background synchronisation. Representatives can capture images and complete audit tasks without interruption; data is automatically synchronised once connectivity is restored.

Serverless Backend

Backend services were implemented using AWS Lambda, providing a scalable, serverless architecture for managing products, tasks, and metadata. The event-driven design scales automatically with field activity and eliminates the need for dedicated server infrastructure.

AI-Powered Shelf Analysis

Captured shelf images are analysed using Amazon Bedrock, leveraging the Claude Sonnet multimodal model. Images are submitted alongside structured prompts that define expected shelf conditions, enabling the system to assess product presence and placement accurately. Each shelf is classified as either compliant (Fixed) or non-compliant (Not Fixed), delivering consistent, repeatable results at scale.

Unified Data Layer

All application data tasks, representatives, store details, product definitions, and images is managed through API layers backed by PostgreSQL. This unified data layer supports both the mobile application and the web portal. Secure data modelling practices and managed cloud services ensure that image data and metadata are stored, processed, and accessed in accordance with security requirements.

The Outcome

The solution replaced the manual WhatsApp and Excel workflow that previously required five or more administrators to reconcile field activity each day. Today, more than 300 field representatives can reliably capture shelf images during store visits, even in low-connectivity environments, without disrupting their existing workflows. 

Backend services process and evaluate shelf conditions in a consistent, repeatable manner, removing subjectivity from audit results and reducing dependence on manual reporting cycles.

Key Outcomes

  • Replaced a 5+ person manual reconciliation process with automated task management and reporting
  • Reduced reporting lag from weekly cycles to near real-time
  • Deployed across 300+ field representatives nationwide
  • AI-powered shelf analysis delivers consistent, objective compliance assessments at scale
  • Offline-first architecture ensures operational continuity in stores with poor mobile coverage

Datafy now has a scalable digital foundation for retail execution, one that can expand alongside their representative network and evolve as AI capabilities continue to advance.

John Fourie, Datafy Managing Director, had this to say, “Working with Silicon Overdrive was an outstanding experience. They took the time to understand our business requirements and delivered a solution that has transformed how we manage and analyze FMCG data, reliable, intuitive, and scalable.

Their technical expertise, attention to detail, and clear communication stood out throughout. They consistently delivered on their promises, adapted to our evolving needs, and proactively suggested improvements that enhanced the final product.

We would confidently recommend them to any business looking for a trusted technology partner capable of turning complex ideas into powerful, user-friendly software.”

Glossary

Offline-First Architecture: An application design approach that allows users to continue working without an active internet connection. Data is stored locally on the device and synchronized automatically once connectivity is restored. 

Serverless Architecture (AWS Lambda): A cloud computing model where backend code runs in response to events without requiring dedicated servers. This enables automatic scaling and reduces infrastructure management overhead. 

Amazon Bedrock: A managed AWS service that provides access to foundation models for building AI-powered applications, including multimodal models capable of processing both text and images. 

Multimodal AI Model: An artificial intelligence model capable of understanding and processing multiple types of data, such as images and text. In this case, it was used to analyze retail shelf images and assess compliance. 

Event-Driven Architecture: A system design pattern where components respond to events (such as image uploads or task updates), enabling scalable and loosely coupled system behavior. 

Data Synchronization: The process of ensuring that locally stored data and central databases remain consistent by automatically updating changes when connectivity becomes available.

Visit Datafy

If you liked this, you’ll love these...

We use cookies to track visitors, measure ads, ad campaign effectiveness and analyze site traffic. We may also share information about your use of our site with 3rd parties. For more info, see, our Cookies Policy, our Privacy Notice. By clicking “Accept All” you agree to the storing of all cookies on your device. In case you don’t choose one of these options and use our website, we will treat it as if you have accepted all cookies.

Accept All

We use cookies to track visitors, measure ads, ad campaign effectiveness and analyze site traffic. We may also share information about your use of our site with 3rd parties. For more info, see, our Cookies Policy, our Privacy Notice. By clicking “Accept All” you agree to the storing of all cookies on your device. In case you don’t choose one of these options and use our website, we will treat it as if you have accepted all cookies.

Accept All