Smartcancy

Inspiration

E-commerce post-order cancellations are traditionally high-friction and blunt: orders are either outright canceled with lost revenue, or support teams are overwhelmed trying to negotiate retention manually. We built Smartcancy to transform the cancellation moment into an intelligent, auditable retention and recovery engine—evaluating candidate recovery options deterministically while keeping humans in the loop for critical approvals.

What it does

Smartcancy orchestrates end-to-end post-order cancellation workflows and revenue protection:

• Cancellation Intake: Ingests incoming cancellation requests seamlessly and links them directly against existing orders.
• Recovery Evaluation Engine: Analyzes each request against six deterministic candidate actions (e.g., discounts, delivery adjustments, substitutions, retention incentives) to recommend the highest-value resolution.
• Human-in-the-Loop Review: Provides operators with clear, explainable decision recommendations that can be swiftly approved or rejected before execution.
• Execution Simulation & Audit Trail: Safely logs simulated execution outcomes and an immutable impact ledger without triggering premature or unintended real-world side effects.
• Operational Reporting & Metrics: Surfaced analytics covering cancellation trends, rule-evaluation breakdowns, and recorded financial impact metrics.
• Cloud-Native AWS Foundation: Architected with AWS CDK for reliable, repeatable infrastructure provisioning with staged rollout guidance targeted for the Sydney (ap-southeast-2) region.

How we built it

• Cloud Architecture & IaC: AWS Cloud Development Kit (AWS CDK) to manage and deploy serverless resources, state workflows, and data stores with reproducible, environment-staged rollouts.
• Decision Pipeline: Deterministic rule engine evaluating incoming order and cancellation payloads against candidate recovery actions.
• Auditing & Simulation Ledger: Event-driven transaction logging ensuring every simulated action and human review intervention produces an inspectable audit trace.
• Operator Interface: Streamlined dashboard designed for swift review, approval workflows, and operational metrics visibility.

Challenges we faced

• Balancing Automation with Control: Ensuring the deterministic rule evaluation could suggest viable recovery actions without taking destructive real-world actions without proper authorization. Building a robust simulation ledger solved this by providing full operational transparency prior to production execution.
• Infrastructure Modularization: Structuring the AWS CDK stacks for seamless multi-stage deployments while maintaining regional compliance and configuration parity for the Sydney deployment target.

Accomplishments that we're proud of

• Implementing a deterministic multi-action evaluation pipeline that assesses six distinct recovery options consistently.
• Creating a zero-side-effect execution simulator that captures every state transition, audit event, and ledger impact accurately.
• Delivering fully reproducible, code-defined cloud infrastructure with automated deployment stages.

What we learned

• Best practices for designing dry-run / simulation layers in transactional retail workflows.
• Structuring clean infrastructure-as-code patterns using AWS CDK for region-specific deployment boundaries.
• Designing human-in-the-loop UX patterns that reduce operator cognitive load when handling edge-case cancellations.

What's next for Smartcancy

• Integrating live downstream connectors for warehouse management systems (WMS) and payment gateways to move from simulated execution to automated execution on approved actions.
• Expanding dynamic candidate actions using predictive customer lifetime value (LTV) models alongside deterministic guardrails.
• Enhancing real-time alerting and anomaly detection across high-frequency cancellation spikes.

Built With

• amazon-web-services
• aws-cdk
• aws-cognito
• event-driven-architecture
• lambda
• python
• serverless
• typescript