Preview Drafting responses with embedded AI

Use the response generator in Sustainability Reporting Manager to automatically create draft responses to disclosure questions. A large language model (LLM) generates a response from your existing policies, sustainability reports, and other documents.

Before you begin

Important: The response generator preview has ended.

Thank you for your participation and feedback, which will inform the next iteration of this capability. For more information or to share additional feedback, contact the product team.

Watch the following short video to get an overview of response generator:


Upload documents at the organization, disclosure, or response level to improve response quality:
  • To upload documents at the organization level, go to Profile > Admin > AI source files.
  • To upload documents at the disclosure level, open the disclosure and go to Actions > View AI source files.
  • To upload documents at the response level, in the disclosure open the question, go to the Response summary field, and click View AI source files.

About this task

Response generator works with narrative-based frameworks such as ESRS Omnibus, GRI, SASB, TCFD, TNFD, UN SDGs, and materiality frameworks.

For more information about response generator, including a full list of the supported frameworks, see About the response generator preview.

Procedure

  1. Open the question that you want to generate a draft response for.
  2. Click Generate draft.
  3. Click the View reasoning link to view the source documents and page numbers that the model used to draft the response. You can also see a quality score out of 10.

    If the supporting documents are insufficient to produce a high-quality response, you receive clear guidance on how to improve them.

  4. Optional: Edit the response.
    You can edit the response directly in the Response summary field. You can also revert to a previously generated response in the history log.

What to do next

AI-generated responses are clearly labeled in the history log. You can recover older versions of responses and their reasoning at any time.