MENA Food Market is a B2B matchmaking platform connecting verified food industry businesses across the Middle East and North Africa. Built on AWS serverless architecture with Amazon Bedrock for AI-powered readiness reports, the platform delivers secure onboarding, document verification, and intelligent directory services.
The Challenge
The MENA food and FMCG industry faces significant challenges in B2B trade partnerships: lack of transparency, inconsistent data quality, and difficulty finding reliable, verified trade partners. Businesses struggle to assess the credibility, operational capability, and trade preparedness of potential partners. Without a trusted, standardised directory, buyers, suppliers, distributors, service providers, and investors waste time evaluating unqualified companies, leading to failed partnerships and missed opportunities. The absence of a unified verification and readiness evaluation framework creates friction across the regional supply chain.
The Solution
We implemented a secure, cloud-native B2B matchmaking platform on AWS. The Company Registration and Verification Engine uses AWS Lambda for data validation and Amazon S3 with KMS encryption for secure document storage. Structured profile data is stored in Amazon RDS, with verification events logged in Amazon DynamoDB for audit trails.
The Directory and Categorisation System structures businesses by category, sector, and readiness level, with Amazon CloudFront enabling global content delivery and optional Amazon OpenSearch Service for semantic searching. The Readiness Evaluation Framework uses AWS Lambda to calculate readiness scores from questionnaire responses, then Amazon Bedrock generates AI-powered readiness reports summarising strengths, gaps, and recommendations.
The frontend is hosted on Amazon S3 and delivered via CloudFront, with Amazon Cognito providing authentication with MFA and role-based permissions. Amazon SES and SNS handle automated notifications, while Amazon CloudWatch, CloudTrail, WAF, and AWS Config ensure monitoring, security, and compliance governance. Amazon OpenSearch Service serves as the platform’s semantic search engine and vector database. Company profiles, product catalogues, and compliance documents are indexed for fast directory search with advanced filtering.
Additionally, OpenSearch stores vector embeddings of trade standards and regulatory documents, enabling the Amazon Bedrock RAG workflow to retrieve contextually relevant compliance information when generating AI-powered readiness reports.
The Results
- Fully functional verified company directory launched, enabling businesses to register, submit documents, and appear with accurate categorisation.
- End-to-end registration and verification workflow operating smoothly without manual intervention.
- Readiness Evaluation Framework reliably assigning readiness levels based on defined criteria, with AI-generated reports delivered via Amazon Bedrock.
- Admin dashboard enabling efficient management of company data, document validation, readiness scoring, and visibility control.
- Stable platform performance with reliable search, filtering, and navigation supporting initial onboarding and early matchmaking use cases.
Lessons Learned
- Implementing the multi-level Readiness Evaluation Framework required careful schema design in Amazon RDS to support flexible scoring criteria across different business categories and sectors. Normalising the categorisation schema early prevented costly refactoring later.
- Using Amazon Bedrock for AI-generated readiness reports required structured prompt engineering to produce consistent, actionable outputs. Balancing report detail with generation speed led to optimising prompt templates for clarity and brevity.
- Document verification workflows benefited from pre-signed S3 URLs generated by Lambda, enabling secure direct uploads without exposing backend infrastructure. Implementing metadata logging in DynamoDB from the start ensured complete audit traceability.
- Deploying Amazon Cognito with MFA and role-based permissions was essential for separating end-user and administrator access, particularly for sensitive document validation and readiness scoring functions.
