AI summaries use generative AI to create concise overviews of a proposal’s purpose, funding needs, and intended impact. This feature helps reviewers save time and focus on strategic evaluation by reducing the need to scan lengthy applications manually.
Enabling or disabling AI summaries
This feature is turned on by default for most clients. Administrators can manage access to AI-generated summaries for new applications.
To turn the feature on or off:
- Select Settings > System > AI Features.
If you do not see this option, you may need additional permissions and should reach out to the Benevity support team for assistance. - Find the AI Grant Summaries setting.
- Use the toggle to enable or disable the feature.
Note: When this feature is enabled, applicants see a consent message in the portal: “Reviewers may use AI to assist in their application review process.” Only proposals submitted after the feature is turned on are eligible for summarization. Older submissions are not included, as they do not contain the required consent.
Generating an AI summary
Once enabled, reviewers can generate an AI summary to quickly assess whether an application aligns with funding goals. Summaries are available only in English.
- Open the proposal you want to review.
- Select Generate AI Summary at the top of the proposal.
- Review the summary, which highlights the proposal’s Purpose, Funding Needs, and Intended Impact.
If the Generate AI Summary button does not appear, the proposal may have been submitted before the feature was enabled and is not eligible for summarization.
Once a summary is generated, it remains visible unless the proposal is updated. If any changes are made to the proposal content, you will be prompted to regenerate the summary. There is no limit to how many times a summary can be generated.
Giving feedback on summaries
You can provide feedback on the accuracy of an AI-generated summary.
- Select the Thumbs Up icon if the summary accurately reflects the content of the application.
- Select the Thumbs Down icon if the summary is inaccurate. You will be prompted to describe what was incorrect.
Feedback is collected and reviewed manually to identify patterns or recurring issues. This helps determine whether the AI model needs adjustments to improve future summaries.
All content is scrubbed of personal information before being processed. Data is never used to train public AI models.