ShappX: serverless commerce with AI support and recommendations grounded in live stock
Morphlix modernised ShappX's e-commerce platform as a serverless architecture on Amazon Web Services (AWS), with customer-support and recommendation agents on Amazon Bedrock AgentCore that cut first-response time for supported questions from hours to under a minute.
- Retail
- Commerce
- Generative & Agentic AI on AWS
- Studio build
- Industry
- Retail · Commerce
- Engagement
- Studio build
Results
< 1 min
first response for supported questions
Down from several hours at peak.
80%
of support cases resolved without a person
Cases handled by the assistant, resolved without escalation.
Higher
recommendation click-through
Than the previous static "related products" module.
3
connected commerce experiences
Storefront, back office and in-store Till.
0
always-on application servers
The challenge
Scale support and checkout without scaling headcount
ShappX sells jerseys, sneakers, tracksuits, caps, bags and accessories online and in-store. It needed to scale checkout, catalogue capacity and customer support without a matching rise in headcount or infrastructure cost.
The same questions, again and again
Support staff answered the same order-status and product questions repeatedly.
Slow first responses
First responses took several hours at peak.
Static recommendations
A static, rule-based "related products" module didn't reflect what shoppers were looking for.
The solution
Serverless commerce, with agents grounded in live stock
A Next.js storefront, a merchant back office and a Till for counter sales, on one backend and one inventory, designed to prevent online and in-store sales from overselling.
AWS Lambda behind Amazon API Gateway, with catalogue and order data in Amazon Aurora PostgreSQL (Serverless) through Amazon RDS Proxy, which manages database connections under Lambda's bursty traffic.
The serverless design scales down between shopping peaks.
The AI layer
Two agents on Amazon Bedrock AgentCore
Both agents act through a defined set of tools (look up an order, search the catalogue, recommend products, hand over to a person) and ground answers in live stock and pricing.
Customer-support assistant
Answers order-status and product questions, and hands over to a person when it is unsure or when asked.
Recommendation agent
Suggests products from live catalogue and order data.
Guardrails and handover
Amazon Bedrock Guardrails filter out-of-scope and unsafe content, and order changes and refunds are handed to a person.
Outcome
Faster answers, with people for the hard cases
Under one minute to first response for supported questions, down from several hours at peak., 80% of support cases handled by the assistant resolved without escalation to a person., Higher click-through on AI recommendations than on the previous static module. and Spend follows traffic: the serverless design scales down between shopping peaks.
Architecture
Serverless on AWS, across two Availability Zones
Lambda functions and the database run in private subnets across two Availability Zones. Infrastructure is deployed as code, with load testing before peak shopping periods.
Edge
Amazon Route 53, AWS WAF, Amazon CloudFront and AWS Certificate Manager, with the storefront on AWS Amplify and assets in Amazon S3.
Compute
AWS Lambda behind Amazon API Gateway.
Data
Amazon Aurora PostgreSQL (Serverless) through Amazon RDS Proxy.
AI
Amazon Bedrock, Amazon Bedrock AgentCore and Amazon Bedrock Guardrails.
Events and operations
Amazon SNS, Amazon SQS and Amazon EventBridge, with AWS Secrets Manager, AWS X-Ray and Amazon CloudWatch.
Technology stack
AWS services behind ShappX
Edge and delivery
- Amazon Route 53
- AWS WAF
- Amazon CloudFront
- AWS Certificate Manager
- AWS Amplify
- Amazon S3
Compute
- Amazon API Gateway
- AWS Lambda
Data
- Amazon Aurora PostgreSQL (Serverless)
- Amazon RDS Proxy
AI
- Amazon Bedrock
- Amazon Bedrock AgentCore
- Amazon Bedrock Guardrails
Events and operations
- Amazon SNS
- Amazon SQS
- Amazon EventBridge
- AWS Secrets Manager
- AWS X-Ray
- Amazon CloudWatch
More from the library.
All case studiesAI · Productivity
Seshira: AI-native project management
An engineering assistant on Amazon Bedrock AgentCore that plans sprints, turns PRDs into tasks and flags blockers, with about 80% less planning effort.
Agriculture, Engineering & Automotive
Sallis: one digital front door for three divisions
One website and a Morphlix Connect enquiry workflow for an agriculture, engineering and automotive company, with 99.8% availability.
Work with the studio
Ready to future-proof
your infrastructure?
Tell us what you are building and where the current stack is holding you back. We will come back with an architecture and a plan.