How Do I Build a Donor Stewardship Tracker With an AI Agent?

If you run development for a nonprofit, the record of who was thanked, who was asked and who has gone quiet probably sits in a few inboxes and a few memories. Below is the order I'd build a tracker in. In the session I built a Chief of Staff AI agent, not a donor tracker, so the steps apply the same moves to a donor sheet. I'm making no claim about results.

What do I give the AI agent first?

A goal, in my own words, and a deliverable. When I set up the Chief of Staff AI agent, this is how I stated the goal:

From the recording, 12:02. And my quarterly goals are to maintain great relationships with the Women Applying AI community. So I'd like to follow up on Slack messages, emails, and send useful information through Slack.

Then I asked for something I could look at: a dashboard, so I could see what needs action right away and what I had let slip through the cracks (12:27).

For a donor tracker, the same two moves take your slots. Example, replace with yours: "My goals this year are to thank every donor after a gift, to follow up on every pledge, and to see which donors have not heard from us. I want a sheet that shows me who needs a touch right now."

How do I ask for the tracker skill?

I told the group the dictate feature is the best way to create skills (36:49), and that I start off by saying I want to create two skills for my Chief of Staff AI agent (36:56). Then I add one sentence, and I typed it on screen for the group:

From the recording, 37:10. Ask me three or four questions to customize my agent's skills.

The steps, as shown October 7, 2026:

1. 01

   Open the chat of your AI agent. I said the dictate feature is the best way to create skills (36:49); on screen I typed the request.

2. 02

   Say or type the request. Example, replace with yours: "I want to create a skill for a donor stewardship tracker. Ask me three or four questions to customize it."

3. 03

   Answer the questions the AI agent asks. Say what you do today: who thanks a donor, who follows up on a pledge, what counts as quiet.

4. 04

   Wait while it thinks. In the session it finished with this:

From the recording, 40:41. It's created two skills for me, an end of day sweep as well as a relationship tracker.

What goes in the sheet?

In my published piece on automating donor pipelines, the Sprint Philanthropy AI agent keeps campaign state in one Google Sheet with four columns. That's the design I use here.

| Column | What goes in it | | --- | --- | | Donor name | The name, or a made-up name while you practice | | Status | Plain words: not yet contacted, contacted, pledged, gave | | Amount | The gift or the pledge | | Last contact date | The day of the last thank-you, call or note |

The same piece adds one line per donor on why they gave, and it lists four jobs for the AI agent: hold the state, report where you stand, draft the next touch, and stop. The major donor side is in my R4 piece on protecting core relationships.

Which donors go in first?

Made-up ones. When I asked my AI agent for a spreadsheet during the session, I asked for demo names:

From the recording, 54:37. Let's see with demo names for a test with fake emails and phone numbers.

Fill the four columns with invented donors, ask for the report, and see whether it reads like something you could act on. Swap in real names only after you've read the skill.

What is the rule for sending?

Mine drafts and doesn't send. My rule for the Sprint Philanthropy AI agent is "You draft. I decide." (I wrote about it in nonprofit AI governance.) In the session I said I give my AI agents guardrails, and that I tell them what they can do and what they can't do (41:26). Write that into the skill: every message is a draft, and a person says yes before anything leaves. That's a written rule. The setting that restricts sending is in tool management, where I showed how to give or restrict permission to send emails (44:19).

What goes wrong

• A skill you haven't read. I said I am very hesitant to copy a skill I can't read first (34:47). The skill your AI agent writes is yours to read too. I showed you can click into the AI agent's skills and see step by step what each one says (41:14). Read it before you rely on it.
• Sharing too much about the people you work with. When I share my own systems, I run a demo so I'm not exposing personal information about the people I work with (55:04). Donors are those people.

What the recording doesn't show: the text of the two skills, a donor sheet, or any result from running one.

Where this comes from

This article grows from the session I taught live with Women Applying AI on October 7, 2026. The quoted passages are my own words, taken verbatim from the recording. The steps, tables and checklists around them were drafted by my team's AI agents from those words and my published pages, and checked against the recording before publishing.

Build Your First AI Agent in One Hour: the full session with the clips, the prompts and the transcript.

Related reading: How to Automate Donor Pipelines & Sustain Giving Across Leadership Changes, AI Major Donor Strategy: Using the R4 Framework to Protect Core Relationships, How Do You Create Skills for an AI Agent?.