1. Case Overview

This case study demonstrates how Climind is used to efficiently extract qualitative and quantitative metrics from numerous ESG reports. By automating data processing and standardization, the customer achieved comprehensive management of ESG data, improving the quality of reports and communication both internally and externally.

2. Customer Profile

Company Name: Illuminem (The world’s leading platform for sustainability information and data. )

Official website：https://illuminem.com/

Industry: Sustainability services and information

Project Goal: To utilize generative AI for extracting and organizing qualitative and quantitative data from ESG reports, enhancing data quality and communication.

It empowers its 500,000+ users to

• gain insights from the world’s largest expert network in sustainability , bringing together 1,500+ experts ( including CEOs of Siemens, Greenpeace, Harvard professors, the UN Under-Secretary-General, Climate Ministers, and more).
• stay updated with unlimited industry news
• maximizing sustainability sales based on ESG insights & proprietary intent data
• monitor & compare the ESG performance of thousands of companies.

It has been awarded the second most beloved startup in Italy by StartupItalia, one of AngelList’s Top Startups in Europe and one of TechStars' Top 12 sustainability startups globally. Also the Italian Ministry of Environment recognized it as an "exemplary and courageous model to follow."Illuminem is the largest global network of sustainability experts, bringing together over 1,5000 global experts, including climate ministers and international CEOs, committed to advancing sustainability. Illuminem provides detailed sustainabilityESG informationnewsreports and dataservices through its platform, attracting 5,000 CEOs and achieving 1.3 million user interactions. Illuminem also established the world's first and largest ESG business alliance, connecting sustainability experts from various industries worldwide to promote knowledge sharing and collaboration in ESG. The company manages billions of euros in equity, spanning sectors such as services, technology, and healthcare. Recently, Illuminem was named one of AngelList's "Top Digital Startups in Europe."

3. Project Background and Challenges

• Inefficiency of Manual Processes: Traditional ESG data collection methods are extremely time-consuming and labor-intensive, with analysts taking weeks or even months to process large volumes of unstructured data, leading to delayed insights and significant resource expenditure.

• Data Quality and Consistency Issues: Manual data collection often results in inconsistent data quality. The absence of AI support increases the risk of human errors, affecting the reliability of ESG assessments.

• Limited Data Coverage and Standardization: Traditional methods have limited data coverage, making it difficult for analysts to thoroughly review all information, leading to potential oversight of important insights. Furthermore, the lack of standardization in report formats and terminology across different companies and industries makes data comparison challenging, adding to the complexity of analysis.

• Delayed Information and Difficulty Handling Unstructured Data: The slow pace of manual data collection leads to information delays, which affects timely decision-making. At the same time, traditional methods struggle to effectively handle unstructured data, such as PDFs, leading to gaps in analysis and understanding.

• Limited Analysis Depth and Subjectivity Risk: Manual processes restrict the depth of analysis, and complex patterns or relationships may be overlooked. In addition, human analysis introduces the risk of bias, impacting the reliability of results.

• Lack of Real-Time Monitoring Capabilities: Traditional methods cannot achieve real-time or near-real-time ESG monitoring, limiting the organization's ability to respond quickly to emerging trends.

4.  Solution and Implementation Process

Solution

To address the challenges associated with traditional ESG data collection methods—such as time-consuming manual processes, inconsistent data quality, and limited coverage—Climind developed the ESG Extractor. This advanced AI tool automates the extraction of ESG data from numerous reports, significantly enhancing efficiency and accuracy. It also features a centralized database for easy access to ESG information and uses natural language processing to effectively handle unstructured data. The Climind ESG Extractor not only streamlines data collection but also enables real-time monitoring and analytics, allowing organizations to respond swiftly to emerging trends. By transforming how ESG data is collected and analyzed, Climind empowers businesses to make informed decisions that contribute to sustainable practices in the digital economy.

Initial Communication and Framework Definition: We first communicated with the customer to clarify their needs for data extraction and reporting. We customized a framework and graphical format that aligns with the company's background to ensure that all key metrics are appropriately represented in the reports.

Solution Steps:

a. Defining Metrics and Creating a Metric Set：Collaborated with the customer to define key ESG metrics, including qualitative and quantitative indicators, and created a metric set to ensure all relevant content is captured.

b. Prompt Development and Accuracy Testing：Developed prompts for data extraction using generative AI and conducted multiple tests to ensure accurate extraction of the required information.

c. Accuracy Detection, Identification, and Optimization of Key Results： Assessed the AI extraction results to identify key issues and optimize performance, including enhancing table parsing algorithms and contextual indexing capabilities to improve overall accuracy.

d. Manual Data Verification and Cross-Validation with AI Results：A team of experts conducted manual data verification and cross-validated it with AI extraction results to ensure data reliability and accuracy.

e. Handling Special Cases and Customer Feedback：Addressed special cases during extraction and implemented two rounds of customer feedback to further enhance accuracy.

f.  Delivering Data and Online Data Visualization Platform：Delivered the final dataset and provided an online data visualization platform where users can view the original source of each data point and perform indexing.

Data Extraction and Integration:

￮  Application of Generative AI:

▪ Utilized generative AI tools to automatically extract data from thousands of ESG reports in PDF and Word formats. The AI identifies key quantitative metrics (e.g., carbon emissions, energy consumption) and qualitative content (e.g., environmental policies, social responsibility commitments) through natural language processing.

￮  Customized Metric Extraction:

▪  The AI categorized and standardized both qualitative and quantitative metrics according to customer requirements, ensuring that all relevant metrics are included in the report and meet the needs of stakeholders.

Data Quality Review and Standardization:

￮  We conducted a comprehensive quality check of the extracted data to ensure consistency with previous reports and corrected deficiencies during data collection. The generative AI effectively identified anomalies in the data and automatically adjusted formats and units.

Data Visualization and Report Generation:

￮  The generative AI not only automated data extraction but also provided the customer with an online data viewing and indexing platform, enabling users to easily locate and understand each data point's specific content and source.

About Climind

Climind - Mind for Climate is a cutting-edge platform dedicated to addressing climate change through advanced AI technologies, including professional Large Language Models (LLMs) and Retrieval Augmented Generation (RAG). By facilitating interactions via natural language, Climind offers comprehensive climate solutions encompassing research and analysis, ESG and climate disclosure report generation, carbon market pricing and trading, and climate risk assessments.

Key Highlights

•     Global Reach: Climind has a significant global footprint, with an average of 100,000 QA page views per month from users in over 49 countries.

•     Data Processing Capability: Climind’s advanced AI models can process reports and documents up to 300,000 words, providing precise source indexing.

•     Extensive Knowledge Base: The AI models cover 14 industry categories and 7 professional source categories, including green finance, climate policy, CCUS (Carbon Capture, Utilization, and Storage), sustainable disclosure and reporting, thermal production and supply, and renewable energy. Key sources include authoritative institutions like IPCC, IEA, CDP, and the World Bank.