What challenges did Syngenta face before AI implementation?

Seed selection is complex because farmers must weigh multiple factors such as product characteristics, growing environments, and field conditions. Sales representatives needed a faster way to generate personalized seed recommendations for growers.

What was the impact of the AI solution?

Sales representatives were able to generate recommendations with analytical models five times faster. The natural-language interface reduced the learning curve for new users. Syngenta evaluated the agent on 100 Q&A pairs and reported high relevancy, conciseness, and faithfulness.

Source

ExpandedEvidence: Low25/100

Syngenta Cropwise AI: Amazon Bedrock Agents for sales recommendations in agriculture

Use case typeCustomer service automationUpdated Jun 13, 2026

Syngenta developed Cropwise AI, a conversational assistant powered by Amazon Bedrock Agents to help sales representatives and growers across North America with seed product recommendations. The solution uses Amazon Bedrock Knowledge Bases with an OpenSearch vector store to retrieve agronomic information, and an ingestion pipeline with AWS Step Functions and Amazon Textract to extract and structure document content. The system is designed to simplify complex seed selection by combining natural-language interactions with multiple AWS services for orchestration, retrieval, authentication, storage, and monitoring.

Organization: Syngenta
Industry: Agriculture
Location: United States
Published: December 2024

Reported outcomes

Strategic outcomes

New product / capabilityEnabled natural-language seed recommendationsSpeed & agilityAccelerated recommendation generationCustomer experience & trustReduced learning curve for new usersCustomer experience & trustImproved recommendation relevance and trust

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Syngenta
Provider: AWS
Maturity: Unknown
Linked source: AWS Machine Learning Blog

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 5

1Conversational AI assistant
2Decision support
3Knowledge management

Seed selection is complex because farmers must weigh multiple factors such as product characteristics, growing environments, and field conditions.
Sales representatives needed a faster way to generate personalized seed recommendations for growers.

Syngenta built Cropwise AI on Amazon Bedrock Agents to converse with sales reps and growers and provide natural-language seed recommendations.
The implementation uses Amazon Bedrock Knowledge Bases backed by Amazon OpenSearch Service, AWS Step Functions for document-processing orchestration, Amazon Textract for text extraction, AWS Lambda for agent actions, Amazon Cognito for authentication, Amazon S3 for document storage, Amazon DynamoDB for metadata storage, AWS AppSync for real-time data sync, and Amazon CloudWatch for logging and monitoring.
The architecture supports tool/action orchestration and query handling across multiple knowledge and data sources to improve recommendation quality and usability.

Technologies

Amazon Bedrock Agents Amazon Bedrock Knowledge Bases Amazon Textract AWS Step Functions AWS Lambda Amazon Cognito Amazon S3 Amazon DynamoDB AWS AppSync Amazon CloudWatch Amazon OpenSearch Service

Sales representatives were able to generate recommendations with analytical models five times faster.
The natural-language interface reduced the learning curve for new users.
Syngenta evaluated the agent on 100 Q&A pairs and reported high relevancy, conciseness, and faithfulness.

Architecture

The solution is a serverless AWS architecture with an LLM agent layer using Amazon Bedrock Agents. User interactions and backend orchestration are handled through Lambda-based action groups, Amazon Cognito for authentication, AWS AppSync for real-time synchronization, DynamoDB for metadata, and CloudWatch for observability. A separate knowledge-base pipeline uses AWS Step Functions to orchestrate document ingestion, Amazon Textract to extract text, Amazon S3 for storage, DynamoDB for metadata, and Amazon Bedrock Knowledge Bases backed by Amazon OpenSearch Service for retrieval.

Sources & evidence1

Evidence: Low25/100Evidence strength