Episode 75: Building Data-Driven Nonprofits with Kelley Hecht

January 28, 20263 min read

Overview

This conversation features a thoughtful discussion between Nathan and Kelley Hecht on how nonprofits can use AI in practical, responsible ways.

Nathan introduces Kelley and highlights her work at AWS, where she helps nonprofit organizations adopt cloud technology and AI with a strong focus on data, security, and ethics. Kelley shares her journey into AI, explaining how early confusion helped her understand the gap between hype and reality. Together, they explore common misperceptions about AI, stressing that it is not about tools alone, but about clear goals, good data, and strong decision-making.

The conversation emphasizes data as a strategic asset and introduces Amazon’s “working backwards” approach to solving problems. Nathan and Kelley also discuss challenges faced by both small and large nonprofits, the importance of leadership support, and the role of culture in AI adoption. They conclude by reinforcing that AI should strengthen human connection, not replace it.

Cover art for episode 75: Building Data-Driven Nonprofits with Kelley Hecht

Episode highlights

02:22 Kelley's Role and Contributions to Nonprofit AI
07:04 Challenges and Misperceptions in AI Adoption
12:11 AI's Role in Nonprofit Ecosystems
15:00 Data Strategy and AI Integration
20:52 Working Backwards: Starting With the Why
24:16 Moving Beyond "AI as the Answer"
32:59 Data as the Lifeblood of the Organization
41:45 Future of AI in Nonprofit Sector

Tips and tools to implement today

Start with the problem – Be clear on the outcome before choosing any AI tool.
Work backwards – Define success first, then design the solution.
Treat data as an asset – Manage data strategically, not casually.
Fix data basics first – Clean, secure, and govern data before using AI.
Ignore the hype – Choose AI based on real needs, not trends.
Match tools to tasks – Use the correct type of AI for the right problem.
Build leadership support – Ensure senior leaders understand and back AI efforts.
Start small and focused – Pilot AI in one area before scaling.
Be realistic about data volume – AI works best with enough quality data.

Favorite quotes

"I think with anything this new and powerful, and AI is a really tangible innovation that came about quickly, which I think was also different—people could experience it in their day-to-day lives in a way that made the outcome impossible to ignore." – Kelley Hecht
"You could do something so fast; it could source something so quickly for you. I think with that came a lot of hope around the problems AI can solve, and the ways it can be deployed and integrated." – Kelley Hecht
"I think the challenge with AI is that it's incredibly complex. You can get a false positive very easily and very quickly. If you don't understand it, you can also get a false negative by using the wrong type of AI tool for the wrong type of problem." – Kelley Hecht
"I think the complexity and time-consuming nature to understand are probably the things holding up many leaders, and yet without some baseline understanding, I don't know that you can fully harness what AI can do." – Kelley Hecht
"I think that the access piece is different this time around than it was 20 years ago when you were first playing around with the early stages of AI." – Kelley Hecht
"The fear is on both sides: the employee may not bring the use case forward, and the employer is, in many cases, feeling a desire to lock things down for a false sense of security. All of this, I think, is a reaction to the pace and power with which these tools have become available. We're starting to see some balancing there, which I hope will lead to more success and adoption for organizations." – Kelley Hecht
"If we look at our sector, it has always known that information and data are how we want to drive decision-making. I don't think it's new to us that we want to be data-driven in our strategic decisions." – Kelley Hecht
"AI is extraordinary and powerful in whichever form you want to use it, and it's only as good as the data it was trained on." – Kelley Hecht