AI chief of staff for nonprofit executive directors

An AI chief of staff is a draft-and-follow-through operator that runs next to the executive director. It is not a chatbot on the website and it is not a strategy partner. It sits in your inbox, your drive, and your calendar and does the work that turns rough thinking into shipped artifacts — board memos, donor emails, agenda drafts, briefing packets — so the ED signs a clean version instead of writing from a blank page.

The first concrete block is inbox triage and donor-email drafts. The AI reads inbound mail, sorts it into the right buckets, surfaces the three to five emails that warrant the ED’s attention today, and drafts responses to the rest — including donor follow-up threads, funder Q&A, and the steady stream of "circling back" notes that consume three hours a week on their own. The ED reviews the drafted thread, edits one or two lines, and sends. The AI also tracks who has answered whom across the week so nothing stalls in the queue.

The second block is board and leadership prep. The AI assembles the next board packet from the finance report, the program update, the development pipeline, and the carried-over decisions from the last meeting — and produces a one-page briefing the ED can read in five minutes. Ahead of each weekly leadership-team meeting it pulls open decisions, blocked items, and last week’s carries into a single agenda so the meeting starts halfway through.

The third block is follow-through. Decisions made in leadership-team meetings move into a running decision log, each assigned to an owner with a deadline, and the AI reminds each owner on the cadence the ED sets. Anything that has been waiting on the ED for more than 48 hours comes to the top of the next morning brief. The AI is also the first reader on every program report, SOP, or funder attachment before it reaches the ED — so the ED sees a polished version, not a raw one.

The line is the same as it is for a human chief of staff: delegate drafting, integration, and follow-through. Hold onto representation of the organization.

Delegate to the AI: first draft of everything written under the ED’s name. Board memos, donor emails, funder proposals (the AI does the draft and integration; the ED signs off), program reports, hiring scorecards, vendor research, weekly meeting agendas, donor briefings from CRM notes, and the operating dashboard the ED opens every morning. The AI also runs the cross-team decision log, the board packet assembly, the donor-pipeline weekly summary, and the cycle of reminding each function about an open commitment until it ships. It coordinates hiring loops — collecting candidates, drafting panel questions, summarizing interview notes — without making the call.

Keep with the ED: anything where the ED’s name matters more than the draft. The final hire. The signature on the grant commitment above the pre-set threshold. The board chair relationship. Donor-relationship stewardship — the AI can prep the meeting, but the ED is in the room. The external narrative: Op-Ed, keynote, public statement. Anything where the founder or ED is the trust-holder for the funder, the board, or the staff. The AI is the operator that gives the ED time to do these things; the AI is not a substitute for doing them.

A second rule: the AI never sends without the ED’s eyes on it. "Send-ready" is a state, not an action. Delegating the dispatch is what turns a useful draft loop into a reputation risk on a Tuesday morning.

An AI chief of staff works when three conditions hold at once. One: there is a steady stream of incoming writing — donor replies, board prep, weekly briefings, funder Q&A — that the ED is currently doing themselves at low quality under time pressure. Two: the rest of the team produces the raw material — finance reports, program updates, CRM notes — in a place the AI can read. Three: the ED reviews what the AI drafts quickly, instead of letting drafts pile up until they are stale.

It does not work when the org is under two people, when the ED is still the only person producing the raw inputs for every draft, or when the rest of the team is not yet used to writing things down in a place the AI can read. Under those conditions the AI produces low-quality drafts that consume more of the ED’s review time than writing from scratch. The AI also does not work for an ED who treats the draft as final — that is not delegation, that is abdication, and the next donor email will surface it.

The simplest test: after two weeks, the ED should be answering email in a third of the time, board-prep should drop from a day to ninety minutes, and the weekly leadership meeting should run against a clean agenda instead of a verbal recap. If those three numbers move, the AI chief of staff is working. If they do not, the inputs are wrong and the fix is upstream — a writing habit at the team level, a CRM that is actually kept current, or a board cycle that is genuinely scheduled — not "more AI".

At the smallest end — two to five people, no operations lead — the AI sits closest to the ED and replaces what an EA + ops manager would do if the budget covered both. Inbox triage, donor-email drafts, board packet assembly, weekly meeting design, and the cross-team decision log. This is where the time-back number is largest and where the risk of "just an EA" is also highest — the AI has to be shaped to do drafting, not just scheduling.

At the middle — five to ten people, often with a development lead or program director in place — the AI splits time between the ED and one or two other functions. Donor pipeline summaries for the development lead, weekly program digest for the program director, board packet assembly for the ED. The pattern that works: the AI has a named "owner" in each function who reviews the AI’s draft from that function before it goes up. Without a named owner per function the drafts drift and the ED ends up reviewing everything twice.

At ten-plus — when the org has hired a chief of staff, an operations director, or a COO — the AI sits underneath that operator as a force multiplier and is shaped to do what the operator does not have time for. Donor briefing memos, funder attachment review, board attachment scrubbing, vendor evaluations. In this shape the AI is not a substitute for the chief of staff; it is what lets the chief of staff stop drafting and start operating.

The right shape is the one that fits the team that exists today, not the team you would like to have in six months. The most common failure mode is sizing the AI for an org chart that does not yet exist — and ending up with a busy operator draft loop that no one on the current team can review.