Grantmaking Phase

Application Areas

Conditions and Requirements

Examples

Problem Definition

Reviewing reports, research, past grants, and other materials to better understand an issue area, spot gaps, and identify emerging needs or trends.

A clear purpose tied to the question being explored; data that reflects the populations and issues of interest; regular checks to avoid reinforcing blind spots or biased assumptions.

King Baudouin Foundation: Used natural language processing (NLP) to categorize its grant portfolio, surface emerging themes, and identify gaps not visible through manual review.

Prioritization

Grouping similar proposals, summarizing community or stakeholder input, and helping staff compare options against strategic goals.

Human reviewers retain final authority; transparent criteria for comparison; safeguards to detect both algorithmic and human bias; clear communication and appeal routes for applicants.

“La Caixa” Foundation: Piloted NLP-based prescreening to cluster research proposals, while retaining human reviewer control over final decisions.

Strategy Development

Synthesizing insights from historical grants and evaluations, evaluating theories of change, testing alternative scenarios, and monitoring stakeholder sentiment or public discourse.

Alignment with mission and long-term goals; collaboration across teams; clarity about when and how AI is used; governance that builds on existing data and AI policies.

Large US foundation (anonymized): Uses an NLP-based knowledge system to query past grants and evaluations to inform strategy and test assumptions.

Partner Identification

Mapping who is working on similar issues, identifying potential partners, and surfacing organizations that may be overlooked.

Human oversight of AI-suggested matches; deliberate pairing of AI-assisted discovery with field engagement to reach emerging or less visible actors; regular review of models for bias.

Giving Balkans / CiviGraph: Uses AI-assisted network mapping to visualize philanthropic activity and coordination opportunities; 

Altruist League: Applies machine learning to support funder-partner matching.

Grant management

Supporting proposal intake and review by summarizing submissions, checking eligibility, flagging financial or operational risks, and retrieving relevant precedents.

Clear decision criteria; staff remain final decision-makers; strong data privacy protections; careful use of commercial tools.

Patrick J. McGovern Foundation: Uses generative AI to guide applicants and support staff review; Grant Guardian: Summarizes nonprofit financial filings to support due diligence.

Evaluate and Learn

Synthesizing grantee reports, evaluations, and external literature to understand what is working and support learning across programs.

Attention to long-term and complex outcomes; regular updating and documentation; awareness of data and capacity limits.

Wellcome Trust: Uses semantic linking to connect grants with resulting research publications; 

Fondazione AIS: Applied AI to synthesize fragmented impact frameworks into a shared learning model;

Better Society Capital: Tested generative AI to apply impact frameworks across its portfolio.

Cross-Cutting Applications

Supporting everyday work across the organization, including internal synthesis and decision support, organizing documentation and institutional memory, applicant-facing engagement and feedback, inter-foundation knowledge sharing, and communication of findings to different audiences.

Organization-wide guidance on AI use; strong data governance; attention to vendor choice, mission alignment, and environmental footprint; use of automation to enhance staff judgment rather than replace it.

Annenberg Foundation: Implemented organization-wide AI governance, staff training, and internal tools for document management and synthesis.