Generate complete food concepts — including recipes, bills of materials, pricing strategies, and marketing plans — in a single automated workflow using foundation models on Amazon Bedrock.
Overview
This Guidance demonstrates how grocery retailers can accelerate private-label product development from weeks of costly agency work to under two minutes using Amazon Bedrock foundation models for automated concept generation. Claude Sonnet 4 generates market rationale, bills of materials with costings, and competitive pricing recommendations, while Stable Diffusion 3.5 creates packaging mockups targeted at specific consumer personas—both running in parallel through serverless AWS services. The application processes authenticated requests through a containerized Lambda function, storing all generated artifacts including recipes, labels, and export bundles in Amazon S3 for immediate retrieval. You gain faster time-to-shelf for new products, the ability to test multiple concepts at a fraction of traditional costs, quick response to emerging consumer trends, and team-wide scalability without any infrastructure overhead.
Benefits
Accelerate product development from months to hours
Reduce concept costs with generative AI
Eliminate the need for multiple specialized experts by using foundation models to produce packaging mockups, product labels, and go-to-market strategies from a single serverless application.
Scale idea generation without infrastructure overhead
Run text and image generation in parallel on a fully serverless architecture, enabling your team to iterate on product concepts rapidly without managing compute resources.
How it works
These technical details feature an architecture diagram to illustrate how to effectively build this solution. The architecture diagram shows the key components and their interactions, providing an overview of the architecture's structure and functionality step-by-step.
Step 1
The user accesses the application through their browser via Amazon CloudFront, which serves the React-based UI.
Step 2
The user authenticates using Amazon Cognito, returning signed JWT tokens to the client.
Step 3
Authenticated requests for food concept ideations are forwarded to Amazon API Gateway, which provides a managed HTTPS endpoint.
Step 4
The application runs as a docker container on a single AWS Lambda function.
Step 5
For text generation tasks, the application invokes Amazon Bedrock, via AWS Lambda using Claude Sonnet 4.5 from Anthropic, a foundation model with strong reasoning capabilities, for text generation tasks.
Step 6
For image generation tasks, the application invokes Amazon Bedrock using Stable Diffusion 3.5 Large from Stability AI, a foundation model for high-quality image generation, for creating packaging mockups. The text and image generation tasks run in parallel.
Step 7
Generated artifacts, including images, BOMs, recipes, labels and export bundles are stored in an Amazon S3 bucket for retrieval via the application.
Deploy with confidence
Everything you need to launch this Guidance in your account is right here.
Let's make it happen
Ready to deploy? Review the sample code on GitHub for detailed deployment instructions to deploy as-is or customize to fit your needs.