---
title: "From zero-shot forecast to purchase order with Amazon Bedrock AgentCore"
description: "This post presents an inventory automation architecture that moves from zero-shot demand forecasts to validated purchase orders while applying safety stock, minimum order quantities, budgets, and promotional adjustments. It combines Amazon Chronos2, hosted on SageMaker Serverless Inference, with four Claude-powered agents built using the Strands Agents SDK and deployed on Amazon Bedrock AgentCore; deterministic Python tools perform computation and agents handle judgment. Product data and future covariates arrive through CSV files in S3, while business rules remain in JSON, enabling what-if forecasts and onboarding without per-SKU training. In internal testing across 50 SKUs over a four-week horizon, the system reached 12.3% median WAPE, reduced onboarding from 2–3 weeks to under five minutes, and lowered estimated monthly inference cost from about $1,091 to $15. A conditional retry loop surfaces budget violations for user decisions instead of silently truncating orders, with traces and evaluator scores supporting auditability."
---

# From zero-shot forecast to purchase order with Amazon Bedrock AgentCore

[AWS](https://yomu.fyi/company/aws) · Hyunsoo Kim, Ph.D. · Sep 11, 2026

**Type:** Problem & solution

## Summary

This post presents an inventory automation architecture that moves from zero-shot demand forecasts to validated purchase orders while applying safety stock, minimum order quantities, budgets, and promotional adjustments. It combines Amazon Chronos2, hosted on SageMaker Serverless Inference, with four Claude-powered agents built using the Strands Agents SDK and deployed on Amazon Bedrock AgentCore; deterministic Python tools perform computation and agents handle judgment. Product data and future covariates arrive through CSV files in S3, while business rules remain in JSON, enabling what-if forecasts and onboarding without per-SKU training. In internal testing across 50 SKUs over a four-week horizon, the system reached 12.3% median WAPE, reduced onboarding from 2–3 weeks to under five minutes, and lowered estimated monthly inference cost from about $1,091 to $15. A conditional retry loop surfaces budget violations for user decisions instead of silently truncating orders, with traces and evaluator scores supporting auditability.

## Context

Inventory managers must forecast demand across many SKUs and convert those forecasts into purchase orders while handling asymmetric overstock and stock-out costs, supplier lead times, promotions, pricing changes, safety stock, minimum order quantities, and budget constraints. Existing per-SKU forecasting and manual business-rule workflows create substantial operational and auditability burdens.

## Approach / What changed

The architecture stores product data in CSV files on Amazon S3 and business rules in JSON, uses Amazon Chronos2 for zero-shot probabilistic forecasting with covariates, and orchestrates four Claude-powered agents through the Strands Agents SDK and Amazon Bedrock AgentCore. Deterministic Python tools perform calculations and file operations, while conditional retries, observability, and evaluations support recovery and auditability.

## Takeaways

- Chronos2 accepts historical and future covariates through the predict\_df() API, enabling promotion and price what-if scenarios without per-SKU fine-tuning.
- In internal testing across 50 SKUs over a four-week horizon, the system achieved 12.3% median WAPE and averaged 8 seconds of end-to-end latency per SKU, excluding cold start.
- When a budget constraint would cause a shortfall, the Supervisor presents options such as raising the budget, expediting delivery, or delaying a promotion instead of silently truncating the order.

**Tags:** [AI Agents](https://yomu.fyi/topic/ai-agents), [AWS](https://yomu.fyi/topic/aws), [Model Serving](https://yomu.fyi/topic/model-serving), [S3](https://yomu.fyi/topic/s3)

- Source: [AWS](https://aws.amazon.com/blogs/architecture/from-zero-shot-forecast-to-purchase-order-with-amazon-bedrock-agentcore)
- Source URL: https://aws.amazon.com/blogs/architecture/from-zero-shot-forecast-to-purchase-order-with-amazon-bedrock-agentcore
- Ingested by Yomu: 2026-09-11T16:02:36.382Z

[Read original post](https://aws.amazon.com/blogs/architecture/from-zero-shot-forecast-to-purchase-order-with-amazon-bedrock-agentcore)