Most companies' biggest climate impact comes from what they purchase, not what they produce. Upstream emissions (scope 3.1) can account for 70% or more of a company's carbon footprint, but the channels to address them are limited. Companies face an impossible choice between inaccuracy or inefficiency: a broad spend-based estimate of their footprint, or an expensive, time-consuming detailed carbon assessment. Without data that points to their decarbonization opportunities, sustainability teams are stuck.
We built Watershed Product Footprints to change that. Product Footprints is sustainability AI that deconstructs every good purchased by a company into the materials and processes behind it to unlock more accurate upstream emissions measurement and smarter product design and procurement decisions.
Product Footprints solves for three persistent problems in scope 3.1 measurement:
- Speed: Produce an upstream carbon footprint in a matter of minutes, compared to traditional approaches that take months.
- Precision: By decomposing each material into its sub-materials and processes, achieve levels of accuracy in emissions mapping that were previously impossible.
- Actionability: With Product Footprints, sustainability and procurement teams can refine their measurements with primary data and run scenarios to see the impact of different purchasing decisions.
The result is an upstream product footprint that more accurately reflects your business so you can make smarter product and procurement decisions.
How Product Footprints works
Product Footprints uses AI to decompose everything you buy into the materials and processes behind them. For any product—from a pair of jeans to an industrial chemical—the AI traces upstream steps like the emissions of different sub-materials, processes, transportation, and more. After an initial mapping, you can further refine your footprint using primary data and first-hand experience with your product and procurement processes. The AI will cascade those changes across your footprint, powering a far closer match to the reality of your business and the ability to scenario plan different product choices. As a result, you’ll surface more levers for decarbonization in your product and procurement decisions.
Watershed’s approach to sustainability AI
Product Footprints was built with embedded sustainability intelligence by a team of climate scientists working alongside AI engineers, guided by our sustainability AI principles:
- Multi-agent architecture for accuracy: Rather than relying on a single AI system that can suffer from confirmation bias, we use separate agents—one for mapping materials to emissions factors and another for independently evaluating those mappings. This prevents the system from validating its own work and reduces errors.
- Three-layer technical approach: Our system processes data through material enrichment to clarify messy company data, supply chain intelligence that creates digital twins mirroring actual procurement processes, and sustainability intelligence that maps to thousands of specific emissions factors based on region, supplier, and technical specifications.
- Greater transparency: Unlike black-box AI systems, we provide confidence scores, document data sources, show the reasoning behind each decision, and let users override mappings. This transparency addresses the trust issues that have limited AI adoption in sustainability reporting.
- Continuous improvement: The platform learns from user corrections and incorporates primary supplier data to improve accuracy over time, while maintaining the ability for sustainability professionals to refine and validate AI outputs.
You can read more about Watershed’s approach to AI here and through our team’s research paper on Criteria for Credible AI-assisted Carbon Footprinting Systems on ARXIV.
Early customer results
Product Footprints has been in testing by dozens of companies with complex supply chains—manufacturing, automotive, chemicals, and more—for [x months]. Early use cases for Product Footprints include:
- Reductions in the real world and on paper: An automaker had purchased low-carbon steel but did not get credit for the resulting emissions reductions because of the limitations of traditional measurement approaches. Product Footprints correctly identified and modeled the lower emissions of the company’s sustainable steel.
- Effective supplier engagement: A life sciences company with over 50,000 global suppliers used Product Footprints to consolidate data from across 70+ internal databases to prioritize impactful supplier engagement. The company had previously struggled with manual processes that caused delays and data gaps in their sustainability program.
- Sustainable procurement: A third company found that AI-generated footprints revealed differences between suppliers and materials that single emissions factors couldn't capture, showing better routes to sustainable procurement decisions.
Impact, not just speed
For decades, companies have made sustainability and procurement decisions based on incomplete or inaccurate data because existing approaches don't work at scale. Product Footprints shows us that AI built with sustainability intelligence can transform corporate sustainability. It's the first launch for Watershed sustainability AI, a suite of tools built not just to speed up workflows, but to drive real decarbonization.
Accessing Product Footprints
If you’re a Watershed customer, speak with your account manager about getting access to Product Footprints. If you’re not yet a customer, you can learn more and request a demo here.