Details
Nothing to say, yet
Details
Nothing to say, yet
Comment
Nothing to say, yet
ES-GENAI is a tool that uses GPT-4.0 to generate narrative responses for CSRD reports. It accelerates the report writing process and can be done in minutes instead of months. The tool is for Accenture teams and not for client use. You can sign in using Accenture credentials and set up the client by adding client details and relevant team members. Upload input documents like annual reports or sustainability reports. The tool allows you to review and rate the responses, ensuring accuracy. There is also a gap assessment feature to identify areas that need improvement. Once the review is complete, you can download the responses in an editable Word format. Contact the provided contacts for a live demo or any questions. Welcome to the demo of ES-GENAI. This is a GENAI-based tool that accelerates CSRD report writing process using a client's existing ESG documents. Using GPT-4.0, we generate narrative responses for each CSRD KPI individually. What would otherwise take months can be done in just a few minutes using this accelerator. This accelerator is a delivery tool for Accenture to be used by Accenture teams and not to be deployed in the client environment. You can sign in to the portal using your Accenture creds to work on any existing client or create a new client that you want to work for. To set up the client, you can add client details, relevant team members who would be reviewing the responses with you. The next step while setting up the client is to choose the ESRS standards applicable to the client as per their materiality assessment. Next, upload the input documents that are to be used as the knowledge base for this tool. This could be the client's previously published reports such as annual reports, sustainability reports, CDP, TCFD or any non-published documents which might be residing internally with the client. For quantitative KPIs, we can leverage a pre-populated template that our team has designed to collect the most recent data from the client as required and mapped to CSRD KPIs. Next, create a session where you map these documents to the corresponding topics. This functionality allows the LLM to work with the most relevant set of inputs for a particular topic. Once a session is created, you see topic cards for each of the topics selected during the client setup stage. Click on any topic card to begin reviewing the responses for that topic. In this window, you can see the KPI, the CSRD guidance text associated with the KPI and the corresponding response. The AI itself gives a score to the response as per how much information was available for that KPI in the input document. If the score is lower, that would generally mean that not enough information was available in the document to extract something that's meaningful. This tool is designed to involve a human-in-the-loop approach, where after the responses are generated, an Accenture SME from our team is going to manually review them. To facilitate the process, we have provided the option of referring to the sources that have been leveraged for a particular answer. You can click on the cited source, the page number, to read where the response was generated from. This helps in ensuring traceability of the responses. The AI partly picks up sentences just as they are in the report, but a lot of times it also uses its generative capabilities to reword in a way that is applicable for that particular KPI. As part of the review, you can give a user rating to the responses and select the relevant tagging as needed. This is a thorough way of ensuring accuracy of the responses and to capture thoughts on where we see gaps. Additionally, we have also built a gap assessment feature in the tool, and there are two kinds of gap assessments that the tool can demonstrate. One is an AI-based rating. This is based on the AI search score and summarized at overall topic and subtopic levels. This is a quick gap assessment that you can see even before you have done any review manually. The second is the user-based rating, which gets generated as and when you complete the tagging of the responses on your own. Like the previous one, you can see how the responses are looking across different topics, subtopics, and you can also see the comments that you have provided while reviewing the responses. Once the review is completed, you can download the responses in an editable Word format to take the copywriting process to the next level. That's a quick overview of the tool. Please reach out to the contacts listed here for a live demo, any questions, suggestions, or client opportunities. Thank you. Thank you.