New tool aims to aid investor engagement by cutting through misleading and false information.
Fintech firm Iceberg Data Lab’s newly launched tool Barbatus utilises generative artificial intelligence (AI) technology to assist ESG analysts in interpretating ESG-related data to improve benchmarking and assess corporates’ environmental impact.
The AI-powered tool generates real-time, text-based and fully sourced explanations in response to questions on the ESG data of portfolio companies.
Speaking with ESG Investor, Matthieu Maurin, CEO at Iceberg Data Lab, described the tool as an “AI ESG assistant”, targeted at analysts, which aims to assist them to more effectively interpret investee firms ESG-related data, improve benchmarking and enhance their understanding on how corporates can decrease their environmental impact.
Maurin said the tool is capable of understanding and extracting qualitative and quantitative information from unstructured and structured data from publicly available information, such as annual and sustainability reports.
Pierre Olivier-Haye, CTO at Iceberg Data Lab, told ESG Investor that one of the fintech’s main issues with existing generative AI products on the market is the “lack of transparency they offer”.
He said that in designing the tool, the company aimed to ensure that transparency “remained a priority feature” which is “crucial for corporate governance responsibilities” in addition to responding to industry allegations of greenwashing with “substantiated data”.
Means, quantity and pace of regulations are key constraints faced by ESG analysts, according to Maurin, which the new tool aims to mitigate.
He said that many ESG analysts’ teams have a “limited budget” or a “huge coverage” in terms of the amount of data that they need to process.
“This kind of tool enables them to both increase their productivity, as well as improve the accuracy, reliability and access to relevant data,” Maurin added.
Iceberg Data Lab’s develops assessment tools and provides environmental data solutions to financial institutions. The company’s dataset covers more than 2,300 different products and services and has a client base of approximately 40 financial institutions including AXA, Aviva, BNP Paribas, HSBC and Robeco.
“Whilst financial institutions are demonstrating increased interest in ESG issues, reporting frameworks have simultaneously become increasingly complex,” said Maurin, adding that Barbatus aims to respond to these issues by helping clients to analyse complex ESG data more quickly, transparently and effectively, allowing them to better understand the impact of their investments on the environment, whilst directly enabling and accelerating their own transition journeys.
Encouraging engagement
Barbatus has been in its pilot phase since September 2022, with it initially developed for use as an in-house tool before being rolled out to Iceberg Data Lab’s own clients.
“We realised that the needs of our own analysts are the same as those required by our clients, including financial institutions, corporates and consulting firms,” Olivier-Haye said.
More than 60 different financial institutions have used the tool as part of its pilot phase.
Olivier-Haye said that the tool offers the ability to assist financial institutions with their engagement activities with investee firms in their portfolios.
“This tool will help with engagement because financial institutions can ask what the main issues of a sector are,” he said, adding that it will assist in the understanding of a given topic and help inform engagement activities between investors and investee firms.
“Barbatus represents a new generation of generative AI, letting our clients know exactly the page, the paragraph and the source from which our AI assistant has extracted information to answer any question,” said Olivier-Haye.
“It will enable ESG analysts to be way more efficient, to track the supply chain, its greenhouse gas emissions, and to monitor compliance of a company with what it has pledged.”
Users upload files, such as annual reports or datasets, to the Barbatus platform with it then sifting through the information based on a prompt entered by the analyst.
As the tool is analysing data provided to it as opposed to being requested to search sources on its own for data like ChatGPT, Olivier-Haye stressed that Barbatus does not deliver “wrong information”.
“The tool is not going to invent information to make the user happy,” he said. “I’d rather that than having false information, because that would be really misleading and catastrophic for our clients.”
