Welcome to Inside Impact, Temelio's monthly video interview series where we sit down with leaders across the philanthropic space for honest, unfiltered conversations about what's working, what isn't, and where the sector is headed. No sales pitches, no jargon, just real conversations with people doing the work. Each month, we'll be exploring a mix of topics, from technology and operations to broader themes shaping philanthropy today.

This month, we're joined by Rachel Kimber, Managing Director of Full Circle Impact Solutions, where she helps foundations and nonprofits build the infrastructure to use AI effectively, ethically, and in service of their mission. Rachel also co-leads NTEN's Artificial Intelligence Nonprofit Community Group. Her career has moved from data and reporting, to data and impact, to data and AI. Alongside Blanch Vance, she led the Oral and Alternate Reporting Initiative, which promotes values-driven, relationship-centered grant reporting. This past spring, she ran the first cohort of her AI governance program for COOs and senior operating leaders. The next one is scheduled to convene February 2027.

‍

‍

Conversation highlights

Organizations aren't stuck because they lack a tool. They're stuck because they aren't ready.

For Rachel, "stuck" looks like standing in front of your board, being asked what's happening with AI, and still having a question mark on the agenda. After 20 years in global grantmaking, compliance, risk, data, and organizational learning, she sees the real problem as organizational readiness: values alignment, staff capacity, the tech stack, and underneath it all, the state of your data. Her framework for moving from stuck to strategic runs through three stages: alignment (what impact are we trying to create?), stewardship (how do we use our resources responsibly?), and enablement (how do we build the capacity to act, learn, and adapt?).

"Technology rarely solves the problems if we haven't clearly identified where we are trying to get to."

‍

You can't govern what you don't know is happening.

Maya shared data from Temelio's upcoming 2026 State of Foundation Technology report: half of foundation staff who use AI now use it daily, up from 11% in 2024. But nearly two-thirds of foundations have no approved list of AI tools, and more than a third rely on staff using free or personal accounts. Rachel's point is that usage itself is no longer the question. Leaders need to understand what staff are using AI for, with what information, and where it is already shaping workflows and decisions. Even a task that seems low risk, like capturing meeting notes, becomes high risk if sensitive grantee data is involved. Maya added that this gap creates a trust problem too: if a grantee asks whether their data goes into AI tools, someone may confidently say no without knowing what is actually happening internally.

"As an organization, as a COO, as a board, you can't govern what you don't know is happening."

‍

Nonprofits carry the bigger risk, and philanthropy holds the power.

Foundations, most of them backed by endowments, have room to be reflective about AI. Nonprofits don't. They hold highly sensitive data about donors and the communities they serve, and they are turning to AI to offset burnout and capacity constraints. Rachel pointed to a growing power asymmetry: funders are setting the norms, sometimes by asking applicants not to use AI at all. Temelio's data shows this shift clearly. In 2024, 69% of foundations accepted AI use in grant applications. In 2026, that number dropped to about 48%. For a development director writing 40 applications while juggling 15 other jobs, that's a real burden.

"If we're investing in nonprofits and saying AI could help, but we're not providing the training and the infrastructure and some of the governance oversight that they might need, we're doing a disservice to the entire sector."

‍

Instead of banning AI in applications, rethink the application.

Rachel's read on the downward trend is that philanthropy is trying to avoid "robots talking to robots." Her suggestion is to get creative rather than restrictive. What if the conversation started at the site visit, with AI helping capture the discussion, all with consent and disclosure, so a nonprofit doesn't have to spend 40 hours on a proposal? Then pilot it, learn, and adjust, including deciding AI isn't the right fit for a given workflow. Maya described a similar vision: an interview transcript becomes a strong first draft, and the nonprofit's job is to confirm it. Rachel also raised the equity questions that come with any new workflow: who gets to shape how AI is used, whose knowledge counts, and who bears the consequences.

"AI can be a really powerful tool to drive equity. It can also amplify existing inequities."

‍

AI is a highlighter for problems that already exist.

Rachel coined this idea at an Oxford-style debate session at the NTEN conference in Detroit, where she argued that nonprofits won't lose their competitive edge by skipping generative AI. Funders fund outcomes, and people don't join the sector for flashy tools. Her bigger point was that AI reveals and accelerates what is already happening inside an organization. Maya shared a Temelio example: an internal support agent trained on hundreds of hours of client trainings kept giving outdated answers, because the product had changed faster than the training material. The fix was connecting the team's AI tools directly to the code base, a source of truth that stays current.

"If we don't have good data governance, if we have biased data, AI is only going to accelerate. We risk automating dysfunction."

‍

Good governance has three layers, and it rests on your data.

Rachel described governance as three levels. A policy sets the organization's overall stance on AI. Responsible use guidelines define approved tools, what data can go in, and who can do what, ideally set by a cross-departmental committee. Practice guidelines come from staff and spell out what is being automated and where human authority sits. The foundation under all three is clean, well-governed data. She pointed to Latinos for Education as an example of an organization that had permissions, integrations, data governance, and a shared learning repository in place before AI, and is now moving fast because of it.

"AI consumes data. If you don't have good source data, now is the time to really start cleaning that up."

‍

Don't start by writing the policy.

Asked what a foundation leader should do this quarter if they haven't started on AI governance, Rachel's answer was direct: skip the policy for now. Start with a governance snapshot, a lab or workshop to map where AI is already being used across the organization. That includes tools that have added AI features since you adopted them. Talk to end users about what they use, for what purposes, and what information they share.

"Don't assume. You have to sit down and talk to your end users about what they're using, for what purposes, and what information they are sharing."

‍

What gives Rachel hope: leaders are showing up, and the sector still has agency.

Rachel is encouraged to see COOs joining cohorts and treating AI as a leadership responsibility, not just an IT problem. Appetite for collective learning is growing, with PEAK Grantmaking and NTEN hosting strong learning spaces. Most of all, she wants the sector to remember that the path isn't set by technology companies.

"It's a tool. We get to decide, and it's not predetermined by the technology companies who want to tell us how to engage with the tool."

‍

Where to learn more

Rachel's PEAK Grantmaking series has two sessions left in its three-part run. NTEN also hosts an AI for Foundations cohort and a resource hub where sector practitioners share prompt libraries and approaches to equity. Rachel plans to launch her next COO cohort in February 2027. Submit your interest here.

Inside Impact is a monthly video series by Temelio featuring candid conversations with leaders across foundations, corporate giving, and philanthropy tech. New episodes every month at trytemelio.com.