Blending Generative AI and machine learning to assess the environmental performance of 13,000 French companies

Challenge
- • Retail banks need to size the carbon emissions of their lending portfolios — a costly endeavor for small and medium sized businesses.
- • In addition, the client sought to propose actions to reduce its portfolio organizations’ environmental footprint.
- • Given the size of the portfolio, our client needed an automated, scalable solution to help them size their emissions and recommend remediation actions.
Our approach
- • We deployed our Celsius.Carbon solution — powered by a combination of machine learning and generative AI — to estimate the environmental impacts of 13,000 organizations.
- • Each analysis contained an estimation of carbon emissions across the three main scopes of carbon emissions, as well as an analysis of the company’s key dependencies and pressures on biodiversity.
- • Based on the key environmental pressure sources identified, as well as the company’s profile (sector of activity, size), the module provided recommended actions to reduce those impacts.
Outcome
- • Results across the entire portfolio gave the bank an overview of the key pressure hotspots on their portfolio, identifying the most material sectors and companies to engage based on their performance, as well as providing avenues for optimization.
- • These results gave the bankers insights on which financing offers to develop to help their clients implement the recommended actions.

Results

An estimation of carbon footprint across the 3 main scopes of emissions

is made for each company

13,000 accurate, in-depth profiles

of organizations

An estimation of dependencies and pressures and biodiversity

is made for each company

Challenges

Accurately estimating a company’s carbon footprint requires comprehensive information about the company or its sector, such as its website, number of employees, and relevant activities. Our client sought this type of estimate for its entire portfolio of investments, which number in the thousands.

Our approach

We deployed Celsius.Carbon, our machine learning and generative AI solution for carbon data, which enables banks to assess the carbon emissions of companies in which they have investments.

Typically, this kind of assessment is done manually through cluster analysis, focusing on the 10 largest companies or those from a specific sector to deepen sector-specific investments —however, Celsius.Carbon allows financial institutions to evaluate the carbon emissions of a much broader range of organizations by level and scope.

We assessed 13,000 organizations using a machine learning database with carbon reporting data from well-established sources. The output was a thorough overview of each portfolio organization, with precise estimations across the three main scopes of carbon emissions and automated recommendations for each organization’s decarbonization trajectory.

Outcome

This depth of insight allows the client to develop a tailored decarbonization plan, highlighting actions their portfolio organizations can take to reduce their carbon footprint. The client can also propose long-term strategies for sustainable transformation — for example, providing loans to transition from diesel to electric vehicles or to refurbish buildings for better heat insulation. By linking actions to financial products, we empower our financial services clients to make impactful changes.

To know more about our GenAI for Business solution