Agentic Visual UGC Reward and Attribution Plan
Structure an automated workflow chain that verifies customer visual media and issues targeted loyalty rewards.
Use this plan to design an autonomous agent system that inspects user-submitted photos and videos against catalog metadata, validates authenticity, and invokes loyalty API tool calls to distribute rewards.
Role: Lead E-commerce Automation Engineer specializing in agentic visual computing and loyalty workflows.
Context
- Catalog Reference: {{merchant_catalog_id}}
- Media Ingestion Scope: {{ugc_media_types}}
- Reward Gateway: {{loyalty_platform_api}}
- Asset Quality Threshold: {{visual_verification_criteria}}
- Fraud Deterrence Rules: {{anti_fraud_rules}}
- Incentive Structure: {{reward_tier_schedule}}
Task
Design a multi-step agentic execution plan that processes visual UGC, matches featured items against {{merchant_catalog_id}}, verifies quality against {{visual_verification_criteria}}, screens for duplicate submissions under {{anti_fraud_rules}}, and calls {{loyalty_platform_api}} to grant incentives per {{reward_tier_schedule}}.
Method
- Specify the autonomous vision tool parameters to parse submitted {{ugc_media_types}} for item presence, brand safety, and lighting.
- Design the vector search tool call to cross-reference extracted image embeddings with {{merchant_catalog_id}} SKUs.
- Integrate perceptual hashing and reverse image search tool definitions to enforce {{anti_fraud_rules}} against web duplicates.
- Construct the agent evaluation loop assigning quality tiers based on {{visual_verification_criteria}}.
- Detail the parameter formation for the {{loyalty_platform_api}} tool call according to {{reward_tier_schedule}}.
- Formulate the attribution tagging workflow that updates product display widgets with verified customer media.
- Build a rollback strategy for handling failed API invocations or contradictory visual classifications.
Constraints
- MUST define programmatic safeguards preventing duplicate reward grants across shared customer accounts.
- MUST NOT execute {{loyalty_platform_api}} if the visual asset fails reverse lookup anti-fraud checks.
- Tool payloads must contain idempotency keys to avoid double distribution.
- Keep output concise and strictly technical.
Output format
1. Vision Analysis & Validation Pipeline
Tool definitions and image acceptance scoring parameters.
2. Anti-Fraud & Attribution Workflow Chain
Sequential logic for SKU matching, deduplication, and asset tagging.
3. Reward Dispatch & API Payload Model
Exact JSON request contract for {{loyalty_platform_api}}.
4. Recovery & Error Matrix
Handling strategies for network drops, mismatched SKUs, and degraded image quality.
Self-review
- Does the plan enforce all elements of {{anti_fraud_rules}} before reward issuance?
- Are the tool schemas compatible with {{loyalty_platform_api}} and {{merchant_catalog_id}}?
- Are the numbered steps specific to computer vision and agent-driven loyalty processing?
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.