How the Jerome L. Greene Foundation is approaching AI with intention

Doing more good, more thoughtfully

The Jerome L. Greene Foundation (JLGF) exists to enrich the lives of New Yorkers through investment in the arts, education, medicine, and social justice. In 2025, the Foundation undertook a structured assessment to determine how AI tools might strengthen that work without compromising the trust, privacy, and relationships at its core. Guided by technology partner Macktez, JLGF has adopted a governance-first model: deliberate, human-centered, and grounded in genuine organizational need rather than technological novelty.

The question we started with

AI is everywhere right now. It’s in the tools organizations already use, in the headlines, and in conversations across every sector. For a foundation like Jerome L. Greene, which holds sensitive information about grantees, donors, and community partners, the question was never simply “Should we use AI?” It was something more considered: “How do we use it in a way that honors the trust placed in us?”

That framing made all the difference. Rather than chasing capability for its own sake, JLGF approached AI adoption the same way it approaches its philanthropic investments with discipline, care, and a clear theory of impact.

A governance-first foundation

Before any tool was piloted, JLGF established thoughtful guardrails. This wasn’t bureaucratic caution; it was a prerequisite for moving forward with confidence. The Foundation identified three principles that would govern all AI use:

Data stays internal. JLGF configured tools, including Gemini within Google Workspace, to ensure that organizational inputs are not used to train external AI models. Grantee reports, board materials, and donor information remain entirely within the Foundation’s control.

Humans remain the authors. Every AI-generated output, without exception, is treated as a first draft requiring human review. No materials are finalized or sent externally without deliberate staff review and approval. AI assists; it does not decide.

Access is earned, not assumed. AI features are configured to respect existing permission structures. Meeting summaries, for instance, are stored in secure internal locations accessible only to those who would ordinarily have access.

Staff also received a practical “Dos & Don’ts” guide — not a document designed to discourage experimentation, but one that creates shared understanding about what responsible internal use looks like during the pilot period.

Solving real problems, not imaginary ones

JLGF has been deliberate about implementing only what solves a documented need. The initial pilots were chosen because they addressed friction that the staff experienced daily, not because they were impressive demonstrations of what AI could theoretically do.

Four areas were identified for the first phase:

Meeting intelligence

Staff and grants meetings generate significant follow-up work. Using Zoom AI, JLGF is now generating consistent meeting summaries and action items automatically, reducing the time staff spend reconstructing what was discussed and who committed to what. The goal is not to replace human judgment about what matters, but to free up bandwidth for it. In addition, in the absence of a formalized approval structure, AI offered a simpler way of capturing critical decision-making during meetings. 

Drafting assistance

Routine correspondence, such as grant acknowledgments, scheduling reminders, and follow-up notes, consumes more staff time than it should. Using Gemini, JLGF is piloting AI-assisted first drafts for these communications. The result is a cleaner starting point that staff can refine in their own voice, rather than a blank page.

Knowledge retrieval

The Foundation holds a significant body of internal knowledge, grantee reports, grantee histories, and programmatic analyses that staff often need to reference quickly. Through Gemini’s “Ask this folder” capability, staff can query curated internal document collections and receive answers sourced directly from the Foundation’s materials, reducing the time spent manually searching archives.

Desk research & synthesis

The Foundation’s program staff regularly engage with the broader landscape of its core funding areas, arts, education, medicine, and social justice. Using Gemini, JLGF is piloting AI-assisted research to identify relevant knowledge, trends, and developments within those domains. The goal is to support synthesis and thematic analysis, allowing staff to bring richer context to grant-making conversations without adding hours of manual research. Like all other pilots, outputs serve as informed starting points that staff review, verify, and build upon.

Each pilot is evaluated against clear performance benchmarks over a 30-day period. The expectation is simple: demonstrate concrete value before expanding further.

The role of Macktez: a partnership, not a prescription

This work did not happen in isolation. JLGF engaged Macktez, a technology consultancy with deep experience working with mission-driven organizations, to serve as a strategic partner throughout the assessment and implementation process.

Macktez’s role was not to arrive with a predetermined solution. It was to listen first. Through a structured two-week assessment, Macktez mapped the Foundation’s existing technology ecosystem, Google Workspace, Microsoft 365, Salesforce, DocuSign, and identified where AI capability already existed, where gaps remained, and where the highest-impact, lowest-risk opportunities sat.

From that foundation, Macktez developed a 90-day roadmap designed for sustained, intentional growth not a sprint toward adoption, but a measured path toward embedding AI in ways that serve the Foundation’s mission and culture.

Looking ahead, Macktez is also advising on the long-term data infrastructure that makes AI most useful: consolidating shared organizational content into governed, reliable repositories so that AI tools are working from authoritative sources rather than scattered files. This kind of foundational data health is often the unglamorous prerequisite to everything else working well.

What this approach reflects

There is a version of AI adoption that looks energetic but moves carelessly: acquiring tools faster than organizations can absorb them, prioritizing novelty over need, and leaving staff to figure out the rest. JLGF has consciously chosen a different path.

The Foundation’s approach reflects something that should not be unusual: that the way an organization adopts new technology is itself a statement of values. Protecting grantee and partner confidentiality, maintaining human authorship, and measuring impact before scaling are not constraints on innovation; they reflect the same integrity that guides JLGF’s philanthropic work.

For other foundations, nonprofits, and mission-driven organizations navigating similar questions, JLGF’s experience offers a practical model: start with governance, not tools. Solve real problems, not hypothetical ones. Measure before you scale. And find partners who understand that the technology should serve the mission, not the other way around.