Before a meeting or cultivation visit, the ED gets a one-page brief on the prospect: giving history with your org, public philanthropic activity, board roles and business background, with links to every source. AI does the reading and summarizing; a person verifies anything that drives an ask amount.
Tools used
ChatGPT or Claude with web search
Candid (Foundation Directory / GuideStar) and ProPublica Nonprofit Explorer (public 990s)
DonorSearch or iWave (optional wealth screening)
Bloomerang, Little Green Light or DonorPerfect (attach brief to record)
Expected outcome · Typical goal
Typical goal: every major-gift conversation starts with a sourced brief and a sensible ask range, prepared in minutes instead of an afternoon.
Setup notes
Limit research to public, professional information: published gifts, donor-wall listings, nonprofit and corporate board roles, 990s of private foundations they lead. Skip family, health, home and personal social media.
Require a source link for every claim; anything without one is dropped. AI web search still gets people confused, so verify identity (same city, same employer) before trusting a match.
Treat ask ranges as a starting point for discussion, never as fact. A person sets the final number.
Store briefs in the CRM with a 'researched on' date and an access limit; donors can ask what you hold on them.
Donor privacy: send only the fields the AI needs (first name, gift amount/date, designation). Don't paste full donor lists, notes on health or family situations, or payment details into a consumer chatbot; use a business/API plan that doesn't train on your data, and honor any donor request not to be shared or contacted.
Prompt / agent config
Replace the {placeholders} with your own details.
Prepare a prospect brief for {Org} on {full_name}, {city}, {known_employer_or_role}. Use only public, professional sources. Sections: (1) Relationship with us: {crm_history}. (2) Public philanthropy: gifts, donor listings, foundation roles, each with URL. (3) Board and civic roles, with URL. (4) Professional background, with URL. (5) Likely interests that match our programs: {programs}. (6) Suggested next step and a cautious ask range with reasoning. Skip anything about family, health, home value or personal social media. If you cannot confirm a source is the same person, say so and leave it out. No unsourced claims.
Last reviewed October 16, 2026. Tool names and features change; check vendor docs before you build.
Where I set this up
I build this for businesses across the Charlotte metro and nearby NC/SC towns, remote or in person by appointment:
A shared grant calendar tracks every funder's LOI, application and report deadlines, with reminders 30, 14 and 3 days out. When a deadline approaches, AI assembles a first draft from your boilerplate library and the funder's actual questions, so the grant writer starts at editing instead of a blank page.
Every online or offline gift triggers a personalized thank-you email within a day, with the required receipt language built in. Gifts above a threshold you set also create a task for a board member or the ED to call or handwrite a note. AI drafts the personal line; the receipt wording comes from a fixed template.
Each month the workflow pulls donors who gave last year (or some year) but not this year, segments them by last gift and interest, and drafts a short personal note for each segment, with a direct ask to renew. Staff approve the drafts, then they send from the CRM or email tool with a follow-up if there's no response.
Non-profits · medium complexity
One practical AI use case a month, for Charlotte-area businesses.