[SEO Subhead]
This Guidance automates the selection of environmental impact factors (EIFs)—sometimes also referred to as emission factors—using foundation models with retrieval augmented generation (RAG) to scale product carbon footprint assessments. EIFs measure activities into metrics to assess potential environmental impacts, like carbon dioxide equivalent (CO2e). The Guidance takes in activity data descriptions (such as purchase orders) and parameters (such as regions), and returns a ranked list of EIFs with similarity scores. Human annotators, invoked by low similarity scores or random sampling, provide ongoing evaluation, improvement, and corrective annotations.
The Guidance is based on using two algorithms published by Amazon: CaML, an algorithm to automate Economic Input-Output based Life Cycle Assessment (EIOLCA) using semantic text similarity matching and Flamingo, an algorithm for automatically identifies process-based LCA EIFs.
Please note: [Disclaimer]
Architecture Diagram
-
Option 1
-
Option 2
-
Option 1
-
This architecture diagram shows how to scale carbon footprint assessments using a knowledge store made up of Amazon S3 and Amazon Bedrock. For Option 2, open the other tab.
Click to enlarge
Step 1
Collect business activity data and emission factors from appropriate sources.Step 2
Index relevant activity category datasets using knowledge bases for Amazon Bedrock to build vector knowledge stores for searching during mapping.Step 3
Ingest the activity data to Amazon Simple Storage Service (Amazon S3). AWS Glue cleans and formats incoming data for mapping activities to impact factors.Step 4
AWS Step Functions iterates through the business activities. It calls large language models (LLMs) hosted by Amazon Bedrock to match the given activity to the most relevant emission factor category using RAG. Matches are stored in Amazon S3.
Step 5
AWS Glue combines the matched activity and category with its corresponding kg CO2e from relevant emission factor datasets. This combined dataset is registered in AWS Glue Data Catalog.Step 6
Report on mapped emission factor data with Amazon QuickSight or Amazon Athena.Step 1
Collect business activity data and emission factors from appropriate sources.Step 2
Index relevant activity category datasets using knowledge bases for Amazon Bedrock to build vector knowledge stores for searching during mapping.Step 3
Ingest the activity data to Amazon Simple Storage Service (Amazon S3). AWS Glue cleans and formats incoming data for mapping activities to impact factors.Step 4
AWS Step Functions iterates through the business activities. It calls large language models (LLMs) hosted by Amazon Bedrock to match the given activity to the most relevant emission factor category using RAG. Matches are stored in Amazon S3.
Step 5
AWS Glue combines the matched activity and category with its corresponding kg CO2e from relevant emission factor datasets. This combined dataset is registered in AWS Glue Data Catalog.Step 6
Report on mapped emission factor data with Amazon QuickSight or Amazon Athena. -
Option 2
-
This architecture diagram shows how to scale carbon footprint assessments using a knowledge store made up of Amazon OpenSearch Service Neural Search. For Option 1, open the other tab.
Click to enlarge
Step 1
Collect business activity data and emission factors from appropriate sources.Step 2
Use Amazon OpenSearch Service Neural Search feature to convert emission factor query text into vector embeddings, and return close matches.Step 3
Ingest the activity data to Amazon S3. AWS Glue cleans and formats incoming data for mapping activities to impact factors.
Step 4
Step Functions iterates through the business activities. It calls large language models (LLMs) hosted by Amazon Bedrock to match the given activity to the most relevant emission factor category using RAG. Matches are stored in Amazon S3.Step 5
AWS Glue combines the matched activity and category with its corresponding kg CO2e from relevant emission factor datasets. This combined dataset is registered in Data Catalog.Step 6
Report on mapped emission factor data with QuickSight or Athena.Step 1
Collect business activity data and emission factors from appropriate sources.Step 2
Use Amazon OpenSearch Service Neural Search feature to convert emission factor query text into vector embeddings, and return close matches.Step 3
Ingest the activity data to Amazon S3. AWS Glue cleans and formats incoming data for mapping activities to impact factors.
Step 4
Step Functions iterates through the business activities. It calls large language models (LLMs) hosted by Amazon Bedrock to match the given activity to the most relevant emission factor category using RAG. Matches are stored in Amazon S3.Step 5
AWS Glue combines the matched activity and category with its corresponding kg CO2e from relevant emission factor datasets. This combined dataset is registered in Data Catalog.Step 6
Report on mapped emission factor data with QuickSight or Athena.
Get Started
Deploy this Guidance
Well-Architected Pillars
The AWS Well-Architected Framework helps you understand the pros and cons of the decisions you make when building systems in the cloud. The six pillars of the Framework allow you to learn architectural best practices for designing and operating reliable, secure, efficient, cost-effective, and sustainable systems. Using the AWS Well-Architected Tool, available at no charge in the AWS Management Console, you can review your workloads against these best practices by answering a set of questions for each pillar.
The architecture diagram above is an example of a Solution created with Well-Architected best practices in mind. To be fully Well-Architected, you should follow as many Well-Architected best practices as possible.
-
Operational ExcellenceRead the Operational Excellence whitepaper
Amazon CloudWatch provides centralized logging with metrics and alarms across all deployed services to raise alerts for operational anomalies.
-
SecurityRead the Security whitepaper
Resources are protected using AWS Identity and Access Management (IAM) policies and principles. Use least-privilege access and role-based access to grant permissions to operators. Data at rest is encrypted using AWS Key Management Service (KMS). HTTPS endpoints with transport layer security (TLS) provide encryption in transit for service endpoints.
-
ReliabilityRead the Reliability whitepaper
These serverless services automatically adapt to demand changes. Amazon S3 is fully elastic, growing and shrinking as data is added or removed, providing durable cloud storage and industry-leading availability without overprovisioning. Step Functions offers serverless orchestration with built-in error handling for modern applications. As the application runs, it maintains state, tracks workflow steps, and stores an event log passed between components. If networks fail or components hang, the application can resume from the last checkpoint, ensuring a seamless workflow.
-
Performance EfficiencyRead the Performance Efficiency whitepaper
Amazon Bedrock allows developers to choose suitable foundation models from leading AI companies for their use cases. Amazon Bedrock is a fully managed service offering high-performing foundation models through a single API, along with capabilities for building secure, private, and responsible generative AI applications.
AWS Glue enables quick data discovery and preparation for analytics, ML, and application development. AWS Glue is a serverless data integration service, eliminating infrastructure management through automatic provisioning and supporting various data processing frameworks and workloads.
-
Cost OptimizationRead the Cost Optimization whitepaper
This Guidance relies on serverless services such as Amazon S3, Athena, and AWS Glue. These services require no infrastructure setup or management and scale automatically to match demand, ensuring minimum resource utilization. Serverless scaling helps reduce overall resource usage and costs by adjusting to fluctuating demands.
-
SustainabilityRead the Sustainability whitepaper
Step Functions call Amazon Bedrock directly, eliminating the need for dedicated compute to process data. This approach reduces compute usage and, as a result, lowers carbon footprint.
Related Content
[Title]
Disclaimer
The sample code; software libraries; command line tools; proofs of concept; templates; or other related technology (including any of the foregoing that are provided by our personnel) is provided to you as AWS Content under the AWS Customer Agreement, or the relevant written agreement between you and AWS (whichever applies). You should not use this AWS Content in your production accounts, or on production or other critical data. You are responsible for testing, securing, and optimizing the AWS Content, such as sample code, as appropriate for production grade use based on your specific quality control practices and standards. Deploying AWS Content may incur AWS charges for creating or using AWS chargeable resources, such as running Amazon EC2 instances or using Amazon S3 storage.
While this Guidance serves as a starting point to help streamline EIF mapping for Life Cycle Assessment (LCA) experts, the results should be vetted by LCA experts prior to considering specific actions. The user should also supplement the results with appropriate information for greater accuracy. This Guidance uses v1.2 USEEIO database as a sample for demonstration purposes only. We recommend that the users apply the most recent version of the database.
The model was tested on products in the US retail sector and does not cover services. The user assumes responsibility to ensure accuracy of the emission factor mapping and to validate the results. Using this Guidance does not guarantee verified greenhouse gas (GHG) disclosures. AWS does not assume any legal liability or responsibility for any errors or omissions or for the results obtained from the use of this Guidance.