On 5 January 2023, the Corporate Sustainability Reporting Directive (CSRD) entered into force. Companies subject to the CSRD will have to report according to European Sustainability Reporting Standards (ESRS). Together the reporting directive and the standards modernize and strengthen the rules concerning the social and environmental information that companies must report.

CSRD brings a number of changes, some of the biggest being the following:

• Clarifications on Double Materiality Assessment
• Reporting must be done in accordance with European Sustainability Reporting Standards (ESRS)
• Requirement for third-party review
• The report must be part of the management report (not a separate report)
• The report must be presented in a digital format according to a common standard in an upcoming European database.

The Double Materiality Assessment is done to identify the relevant areas to be included in the reporting according to ESRS. It is grounded in the IRO principle, which stands for Impact, Risks, and Opportunities. This principle dictates that analyses should consider both the impact of organizations on people and the environment (the inside-out view) and the influence of sustainability-related developments and events on organizations (the outside-in view). These two views together help identify what aspects need to be reported, based on consequential and financial materiality.

From the inside-out perspective, we examine how organizations affect environmental, social, and governance (ESG) aspects of sustainability. Conversely, the outside-in view focuses on the financial risks and opportunities that arise from sustainability-related developments and events.

It’s crucial to highlight that this new approach necessitates a shift in reporting timelines. We are no longer confined to discussing the present; instead, we must also describe potential future scenarios.

CSRD is about big unstructured data

In large companies, the supply chains can be very long and complex, and the related contracts as well. The CSRD requires that events are reported over the entire supply chain at the ESRS topic level. Understanding the effects of various events on these long and complex structures can be time consuming. An event for an under supplier may escalate up the chain and have an impact on simulations of effects and risk evaluations. With future scenario creation, things quickly become overwhelming.

It is, however, possible to make this task a real-time activity that makes sure the organization is always up to date concerning their environmental and social information. This requires understanding of the information flows and how to best make use of the resources and technologies at hand – such as AI and data.

How can AI help with CSRD?

AI can help organizations with:

• Gap analysis
• Scenario analysis
• Simulation of Effects, and
• Risk evaluation

To begin with let’s focus on one of the aspects, gap analysis, which is the first step organizations should take after the Double Materiality Assessment to ensure an effective implementation of CSRD reporting.

A gap analysis in the context of CSRD is used to assess how an organization needs to change its disclosure practices and information gathering and processing to meet the requirements set by the reporting standard and the directive. To understand how AI can help with this part of the analysis, let’s review the steps of a gap analysis in the CSRD context.

The typical steps of the gap analysis

Start by learning about the current state

The common steps for gap analysis are to first conduct training about the reporting requirements to ensure the entire team understands and is up to date with the reporting standards. For obvious reasons this is one of the most important steps, as without an understanding of the requirements set by the ESRS and CSRD, the team will struggle to effectively identify the gaps.

The next step is to choose a tool where the results of the analysis are collected. Many organizations start with an Excel spreadsheet, but there are many dedicated tools available as well.

When a tool has been selected, the actual analysis starts by reviewing the current state of sustainability reporting in the organization. This is done by examining existing sustainability reports and any previously used frameworks, building an inventory of sustainability metrics collected today and the documentation of how the metrics data is collected, and what the current practices are for disclosing the sustainability information.

Explore the standards and identify the gaps

The relevant standards and the information about the current state are explored to identify gaps between current disclosure of sustainability information and the expected level defined by the standards. The identified gaps can be organized into three categories:

• Fully aligned – no identified gaps and all the information is already disclosed.
• Partially aligned – either new or different information needs to be disclosed.
• Not aligned – information is not disclosed today and should be added to the report.

Once the gaps have been categorized, a roadmap with a plan of execution and prioritization can be created to close the gaps. If the available timeline is tight, some gaps might need stopgaps to meet the immediate requirements while the long-term information management processes are built.

After that, it is of course time to go and apply the plan, and finally the organization should keep perfecting the established information processes to keep the quality of the reported information high over time.

AI supporting the gap analysis

Assistant analyzing the gathered information

The amount of information handled during the gap analysis can be large. Generative AI together with document understanding can help when the current state is analyzed, and when identifying potential gaps. While human supervision is still recommended, AI can reduce the amount of work during the gap analysis substantially by identifying relevant documents and pointing to where the information can be found.

Data gathering

In some cases, closing the gaps identified during the analysis will require requesting and finding more information, for example about a supplier. Based on our practical experience, generative AI supported data scraping can very effectively gather a broad range of information about a topic of interest and prepare gathered information for further processing. After the gap analysis this can be assigned as a continuous process, notifying the key roles in your organization about sustainability events or developments that can affect the CSRD reporting.

Managing the process of gathering data

An organization-wide effort for a large organization will require coordination and regular follow up. Investment in these required resources can cut the budget available for other initiatives. Based on our experience, AI can support in coordination of tasks and processes, reducing the need for human effort in such administrative tasks.

Breaking the task down to smaller chunks

With all the advancements of AI, you still cannot just “throw AI” to these tasks and expect success. Creating a successful implementation will require planning and collaboration with the relevant stakeholders and technical experts. It is also important to identify the specific tasks where AI can be useful, what type of AI to use for each of them, and break down the task into smaller subtasks to minimize the problem AI needs to solve.

This minimization of the problem at hand for the chosen AI solution is one of the key aspects for a successful AI implementation. Similar to a child being presented with a toy store to choose their prize, AI can struggle to efficiently identify reasonable options if the problem given to it is too large. When the problem is divided up into smaller pieces, the number of options available for each smaller problem decreases to a more manageable level. Similar to a child presented with a smaller and more manageable selection of toys, AI is more likely to perform better when it can focus on a well-defined problem.

ChatGPT gave us so much more than just a chat

Even today, the latest versions of AI are seen as glorified chatbots that help with our day-to-day tasks. But the potential of the technology is far greater. Tasks like identifying relevant content and creating gap analyses for CSRD according to given standards are examples of what AI can help with. There is even more potential with scenario analyses, risk evaluations and simulation of effects, for example through the use of AI agents.

An AI agent is a system that is capable of autonomously performing tasks without human intervention. Such an agent can be created to help monitor the events impacting the CSRD reporting in real time, updating the underlying analyses (and even the scenarios) for each ESRS topic continuously. But we will have to return to these in a future post.

Ultimately, well-executed CSRD reporting will not just be about regulatory reporting. It can become a significant strategic asset for your organization as automated, real-time information gathering and analysis allows your organization to be quickly informed about potential financial risks and opportunities, and act upon them faster than the competition.

If you are struggling with your CSRD reports and your team is getting overwhelmed by the amount of information needed, let us help you use the technology behind the chat that you use for spell-checking to do something much more impactful. Reach out and let us know about your unique situation.