Salesforce plans generative AI boost for ESG reporting with Net Zero Cloud

Generative AI will soon be everywhere — including in Salesforce’s Net Zero Cloud environmental, social, and governance (ESG) reporting tool.

Salesforce will add new features to Net Zero Cloud to automate some aspects of preparing ESG reports — something investors and regulators are increasingly paying attention to — and later upgrading them with new generative AI capabilities.

Net Zero Cloud uses data held within the Salesforce platform to help enterprises report on their carbon footprint and manage other social and governance metrics.

The generative AI capabilities of Einstein for Net Zero Cloud mean it can digest previous annual financial and ESG reports, and current data about greenhouse gas emissions from financial and production systems, said Sunya Norman, Salesforce’s head of ESG strategy and engagement.

“Then, Einstein suggests new report content, without the human having to do anything, that’s purpose-built for the ESG framework you’re trying to report against,” she said.

Salesforce expects to add the new generative AI capabilities in spring 2024, it said.

There are a number of ways that generative AI can help employees process the masses of data involved in ESG reporting, according to Amy Cravens, a research manager contributing to IDC’s sustainable strategies and technologies team.

One of those is deriving insight from data. “The use of generative AI for data queries is one application I’m seeing that can be very impactful in organizations’ sustainability decisioning, identifying which actions or strategies are going to have the biggest impact to sustainability performance,” she said.

Double materiality

Meanwhile, two other upgrades to Net Zero Cloud will become globally available from October 2023.

One is Materiality Assessment — a tool to help determine whether something will have a material effect on a company’s finances or its environmental impact, for example. It will use a customizable questionnaire to help enterprises gather input from stakeholders and score the results to rank the importance of topics.

“This is all about allowing companies to identify and prioritize the ESG topics that are most relevant to their specific business model,” Norman said.

The other is CSRD Report Builder, which will help businesses comply with the requirements of the European Union’s recently introduced Corporate Sustainability Reporting Directive (EU CSRD). This includes the notion of “double materiality” — the effect of the enterprise on people and the environment, and the effect of sustainability matters on the enterprise.

Net Zero Cloud already includes a number of other tools to build reports that follow voluntary reporting rules from the Carbon Disclosure Project (CDP), Global Reporting Initiative (GRI), and Sustainability Accounting Standards Board (SASB).

However, compliance with the EU CSRD will be mandatory. Beginning in 2024, the very largest enterprises will have to report additional sustainability information such as Scope 3 (value chain) emissions, with the legislation’s reporting requirements gradually extending to smaller organizations over the following two years.

Salesforce isn’t the only big software vendor with an eye on the CSRD: Back in May, SAP unveiled its plans to help enterprises build a green ledger to track their carbon balance alongside their financial balance.

According to EcoAct — another technology company with net zero emissions services to sell — the CSRD will extend the coverage of Europe’s emissions reporting regulations from 12,000 companies today to 50,000.

Double-edged

Using generative AI to help enterprises keep tabs on their greenhouse gas emissions, as Salesforce plans to do, can be a double-edged sword, as building and tuning the large language models (LLMs) they run on is energy intensive, and not all data centers use clean energy. In other words, using generative AI can increase greenhouse gas emissions.

One way to avoid that dilemma, Norman said, is to avoid using AI for its own sake, ensuring that we only use AI models that are fit for purpose. “That’s the number one thing we can all be doing as an industry,” she said.

IDC’s Cravens welcomed Salesforce’s focus on fitness for purpose.

A useful question to ask, she said, is, can the use of AI deliver results that human analysis would not generate? “I think in the instance of ESG and sustainability, there are use cases that it absolutely makes sense,” she said.

Norman added it’s not just about clicking (or not) on that generative AI button, though.

“There are a lot of other exciting levers we can pull” to reduce greenhouse gas emissions from AI operations, she said.

Those include using cleaner energy — Salesforce sources 100% clean energy for its global operations, she said — and more efficient hardware in data centers. Nvidia, in particular, has been hammering home the message that GPUs are more efficient for training LLMs.

But amid all this technology, CIOs in search of better energy use should never lose sight of the human factor.

“Developers can produce more efficient code,” said Norman, whether that’s the base code running an application or model, or “even UI designers, who are creating a better UI that’s more efficient and gives a better experience to the end user.”