Services provided
- Artificial Intelligence
- Data & Analytics
- Cloud Architecture
- Supply Chain Optimisation
Platforms used
- AWS (Amazon Bedrock, AWS Step Functions,
AWS Lambda, Amazon ECS Fargate, AWS Batch,
Amazon S3, Aurora PostgreSQL, Amazon API Gateway, AWS WAF, AWS DMS, Amazon Kinesis)
Engagement length
- 8 weeks
Other stats
- 273 active retail stores analysed
- ~$28.3M annual transport spend assessed
- ~360,000 shipment jobs analysed across a
two-month period - 12-stage decision manufacturing
framework applied - ~300,000 decision and API events
estimated per month at production scale
Background on the customer
Our customer operates a large multi-brand retail network with a complex fulfilment and distribution footprint across Australia.
As customer expectations for delivery speed, cost, and reliability continue to evolve, getting fulfilment decisions right has become increasingly difficult. Balancing inventory availability, transport costs, carrier performance, service levels, and customer promises requires more dynamic decision-making than traditional, rules-based routing engines are able to support.
Challenge
The organisation’s fulfilment-routing capability was built around a static, rules-based engine designed for a more predictable operating environment.
While effective for basic routing scenarios, the platform struggled to adapt to changing business conditions and increasingly complex fulfilment requirements.
Key challenges included:
- Limited ability to dynamically optimise carrier and route selection
- Difficulty responding to changing transport rates and carrier availability
- Limited visibility into fulfilment decision rationale
- Inability to easily compare alternative fulfilment scenarios
- Manual intervention required for exceptions and policy changes
- Growing transport expenditure with limited optimisation visibility
- Lack of explainability and auditability across routing decisions
Analysis of fulfilment operations identified meaningful opportunities to optimise transport costs while maintaining customer delivery commitments and service-level objectives.
Stack Highlights
The Approach
Codex partnered with the retailer to design an AWS-native OMS Decision Platform capable of augmenting and progressively replacing static fulfilment-routing rules with explainable, data-driven decision intelligence.
Our approach included:
1. Conducting fulfilment-network discovery and shipment analysis across the retail footprint
2. Evaluating transport cost, carrier selection, and fulfilment-routing opportunities
3. Designing a decision-manufacturing architecture to support optimisation, simulation, and policy-driven decision making
4. Applying the 12-stage decision manufacturing approach covering data ingestion, analysis, simulation, optimisation, policy evaluation, decision construction, and reporting
5. Designing event-driven, multi-scale orchestration using AWS Step Functions, Lambda, ECS Fargate, and AWS Batch
6. Leveraging Amazon Bedrock to generate plain-language explanations for fulfilment decisions and recommendations
7. Establishing immutable decision evidence and audit lineage for governance and compliance
8. Defining shadow-mode deployment patterns to validate optimisation outcomes before production cutover
This approach enables fulfilment decisions to be evaluated against multiple competing objectives, including transport cost, delivery speed, SLA risk, inventory constraints, and carrier performance.
Technical Outputs
Codex delivered a target-state architecture and implementation roadmap for an intelligent fulfilment decision platform.
Key outputs:
The resulting architecture provides a foundation for explainable, auditable fulfilment optimisation at enterprise scale.
QA Platform Highlights
Intelligent Fulfilment Optimisation
Moves beyond static routing rules by evaluating fulfilment decisions across carrier cost, inventory position, SLA commitments, and serviceability constraints.
Explainable Decision Intelligence
Provides transparent, plain-language explanations and evidence for every fulfilment recommendation, increasing operator trust and auditability.
Scenario Modelling and Simulation
Enables alternative fulfilment strategies to be evaluated before execution, helping balance customer outcomes against operational costs.
Immutable Decision Evidence
Every recommendation is supported by sealed evidence, lineage tracking, and replayability to support governance and continuous improvement.
Scalable AWS-Native Architecture
Designed to support hundreds of thousands of fulfilment decisions per month through a combination of serverless, containerised, and batch processing services.
Business and Commercial Outcomes
The discovery engagement established a clear pathway to modernising fulfilment decision-making and identified measurable optimisation opportunities across the retail network.
- Identified opportunities to optimise a transport-spend baseline of approximately $28.3M annually
- Established a roadmap to progressively replace static fulfilment-routing logic with dynamic optimisation
- Defined governance controls for balancing transport cost, customer experience, and SLA commitments
- Reduced implementation risk through shadow-mode validation and phased rollout strategies
- Improved visibility into fulfilment decision drivers and optimisation opportunities
- Created a scalable platform architecture capable of supporting future AI-powered supply-chain capabilities
- Established a foundation for continuous optimisation through simulation, feedback loops, and decision analytics
The platform positions the organisation to make faster, more transparent, and more cost-effective fulfilment decisions while supporting future growth and operational complexity.
Talk to Us
We would love the opportunity to connect and understand more about the problems you are trying to solve.
Adrian Campbell
Chief AI Officer
Martin Campbell
Managing Partner
Get in touch to coordinate a meeting with one of our technical experts.
Australia: +61 7 3132 3002.
