How a nonprofit executive search firm used AI to handle volume — screening, notes, drafting, and board scheduling — so its consultants could spend their time on judgment.
Charity Search Group places chief executives and senior officers at nonprofits. A single search can draw thousands of applicants. Each one deserves a fair look, but only a few dozen belong in front of a client, and only a handful in the finalist round. The work that decides a search — assessing a candidate’s fit with a board, a mission, and a moment in an organization’s life — is human work, and it was being squeezed by everything around it.
Around it: hours of calls with no reliable record, résumé review that ran late into evenings, position profiles and reports drafted from scratch each time, and the logistics of keeping a client’s search committee — sometimes a dozen or more board members with their own calendars and their own questions — informed without meeting on the same topic three times.
The question was not whether AI could do the consultants’ job. It couldn’t, and no one wanted it to. The question was whether it could take the volume off the desk so the judgment had room.
Every applicant is assessed against the position profile by AI before a person sees the pool. The output is not a ranking the firm acts on; it is a shortlist of dozens, with the reasoning attached, that a consultant then reads and vets by hand. Applicants the model was unsure about are flagged for human review rather than dropped. Thousands became dozens without anyone losing a fair look.
Every candidate call, client check-in, and committee meeting is transcribed and summarized automatically, with action items pulled out. Consultants stopped splitting attention between the conversation and the notepad, and the firm gained a searchable record of every search — who said what, when, and what was promised. Summaries are reviewed before they are shared; nothing goes to a client unread.
Administrative and search staff use AI to produce first drafts of position profiles, candidate summaries, committee reports, and client correspondence, working from the firm’s own templates and the call record. The consultant edits and signs off. The blank page disappeared; the review step did not.
Search committees generate a large volume of overlapping questions and availability constraints. AI now groups incoming board-member correspondence by topic, identifies what has already been answered, and proposes meeting slots and agendas that cover each topic once with the right people. Board members meet less often and leave with fewer open items.
AI narrows the pool and explains why. A consultant makes every advance-or-decline decision, and borderline cases route to a person by default.
Nothing leaves the firm without a person reading and sending it. Drafts are drafts until a consultant says otherwise.
Candidate and client information stays in the firm’s own systems under its own access rules. Tools were chosen and configured with that constraint first.
The most visible effect was on where time went. Screening that once consumed days of a consultant’s week became a review of a vetted shortlist. Call notes stopped being a chore that lagged the conversation by a week. Reports that used to start from nothing now start from a draft that already knows what was said.
For clients, the change showed up as fewer meetings that covered old ground, faster turnaround on committee questions, and a more complete record of the search from first call to placement. For candidates, it meant every application was actually assessed, and the people who reached a conversation had earned it on the merits of the profile rather than the luck of where they landed in the pile.
No role was eliminated. The hours came back to the people who were already there and went into the part of the job that clients pay for.
Charity Search Group shares operating leadership with A2 Digital. We practice what we preach: this is our own firm, and these methods ran there before we recommended them to a client. See also the Talent Echo case study.
Thirty minutes to find out where AI takes the first pass in your operation — and where it shouldn’t.
Book a Call