All jobs

Which Donors Gave Last Year but Not This Year? A Fundraising App's Monthly Airtable Check for Each Nonprofit

A fundraising software company makes one agent per nonprofit that reads its Airtable donor base each month and lists lapsed donors and gifts never thanked.

Run this on RushOpens the console with this prompt ready to send. Connect Airtable there first.

What the agent does

This is for a company that sells fundraising software to small nonprofits: donation pages, recurring giving, receipts. Many of its customers keep their donor records in an Airtable base they built themselves, a Donors table and a Gifts table, because it was free and flexible when they had one volunteer and a spreadsheet. Every month the development director wants the same answer: who gave last year and has not given this year, which monthly donors have stopped, and which gifts never got a thank-you. The company saves one agent as a template whose connections name airtable, Airtable's MCP server at https://mcp.airtable.com/mcp, and makes one agent from it for each nonprofit, storing the agent id on the nonprofit's account.

Each nonprofit connects its own Airtable once from the software's settings. The company's backend calls POST /o/{org}/agents/{agent}/connections {"type": "mcp", "name": "airtable"} and opens the authorize_url in a popup. On Airtable's page the nonprofit chooses exactly which bases, apps and workspaces the agent may reach, so a base with board minutes or staff records stays out, and they can change that later from their Airtable account. What the agent can do in a base also follows the signer's Airtable role; an editor's role allows changing records, which is why the template's instructions forbid it.

On a run the agent searches the nonprofit's bases for the one named in the prompt, reads the Donors and Gifts tables' fields, and maps the nonprofit's own names to what it needs: the gift amount and date, the link from a gift to its donor, a recurring flag, the date a thank-you was sent, and the donor's email and mailing address. It reads every record, then works out each donor's last gift, their total in the last fiscal year and in this one, using the fiscal year start the prompt gives.

It lists three groups. Donors who gave in the last fiscal year and nothing so far in this one, largest past total first. Recurring donors whose latest gift is more than 45 days old, which can mean a card that expired. Gifts at or above the amount the prompt names with no thank-you date seven days after the gift. When a field the checks need is missing, it says so once instead of flagging every row. It creates, edits and comments on nothing.

Start it with one API call

POST https://api.phoenix.so/o/{org}/sessions with this body:

{
  "agent": "<agent id>",
  "input": "Find the Airtable base named 'Donors and Gifts'. Read the fields of its Donors and Gifts tables and map: gift amount, gift date, the gift's linked donor, recurring flag, thank-you sent date, donor email, donor mailing address. Read every record of both tables. The fiscal year starts July 1; today is 2026-11-14. Return only JSON with: lapsed: donors with gifts in FY2026 (2025-07-01 to 2026-06-30) and none since 2026-07-01, each {record_id, name, last_gift_date, fy2026_total, email}, largest fy2026_total first; recurring_stopped: donors marked recurring whose latest gift is before 2026-09-30, each {record_id, name, last_gift_date, amount}; not_thanked: gifts of 250 or more dated before 2026-11-07 with no thank-you date, each {record_id, donor, amount, gift_date}; missing_fields: fields you could not map. Do not create, update or delete any record, field, table or comment."
}
  1. Get an API key. In the console, open Settings, then API keys, and create a key for your org. Send it in the Authorization header as a Bearer token.
  2. Send the request. POST the body in request.json to https://api.phoenix.so/o/{org}/sessions, with your repository as the resources URL. The run starts in its own cloud sandbox.
  3. Read what came back. Follow the run on its live stream or wait for the run.completed webhook, then read the result and its trajectory.

What comes back

The answer is JSON in the session's result and in the run.completed webhook. The fundraising software shows it to the development director as three short lists, each line linking to the record in their Airtable base, alongside the donations its own pages took that month.

The director checks each list before acting on it. A lapsed donor may have given this year by cheque that nobody has entered yet; a stopped recurring donor may have switched to giving through a workplace scheme; a gift with no thank-you date may have been thanked in person. Who gets a call, a letter or nothing is the director's decision, and the thank-you itself goes out from the nonprofit, not from the agent.

The run reads donors' names, gifts and contact details, and the session's transcript holds what the tools returned, so the software company treats each nonprofit's sessions as that nonprofit's donor data.

Run it on every month, per nonprofit

On the first business day of the month the company's scheduler sends one POST /o/{org}/sessions per connected nonprofit, with that nonprofit's agent id as agent, its base name, fiscal year start and thank-you threshold in input, and a request_id that is a UUID v5 of the nonprofit's id and the month so a retried job starts no second run.

A director can also press a button before a board meeting or a year-end appeal, which sends the same request at once. Each answer arrives on the key's webhook, signed per Standard Webhooks, and the handler matches execution_id to the nonprofit it saved at start.

The use case this job follows, with its automation and trigger: Run an agent per customer task.

Why run it on Rush?

No two nonprofits built their base alike: one tracks gifts in the Donors table as a rollup, another has a Pledges table, a third calls the thank-you field "Ack sent". A fixed integration needs a mapping per customer and breaks when a volunteer renames a field. The agent reads the fields every run and maps them, and says plainly when it cannot. Each nonprofit's Airtable belongs to its own agent, so a run for one never reaches another's donors, and the nonprofit chose the bases it shares on Airtable's own page.

The software company keeps one template. Changing a rule, such as the 45-day window for monthly donors, is one PATCH to the template that reaches every nonprofit's agent on its next run, and GET /o/{org}/sessions?agent=<agent id> lists one nonprofit's runs with their total_cost_usd.

Frequently asked questions

Can the agent see all of a nonprofit's Airtable?

Only the bases, apps and workspaces the nonprofit picked on Airtable's sign-in page. They can change the selection later in their Airtable account, and the next run sees the new set.

Will it write the thank-you letters?

Not in this setup. A second prompt could draft letters from the not-thanked list for the director to edit, but a thank-you carries the nonprofit's own voice, so a person writes or approves every one.

Does the nonprofit need a Rush account?

No. They sign in to Airtable only. The agent, its runs and the bill belong to the person who created the software company's API key.

What if the nonprofit moves its donor records out of Airtable?

Then the base search finds nothing and the agent says so. The company can delete that nonprofit's agent with DELETE /o/{org}/agents/{agent}, which also disconnects its Airtable.

More Airtable jobs

More jobs like this: Run an agent per customer task

Keep reading