22. AI Spotlight
By the end you'll be able to
- Use AI to summarize a 990-PF, extract foundation priorities, and draft a funder profile.
- Generate keyword synonym clusters and a first-cut NOFO compliance matrix with AI assistance.
- Identify the three to four common hallucinations that long AI outputs produce in research work.
- Apply Human-in-the-Loop verification to every AI-generated research artifact.
AI accelerates prospect research without replacing the judgment that makes it valuable. The right way to use AI in Week 2 is as a fast first pass: summarize a 990, extract a foundation's stated priorities from its website, draft an initial funder profile, generate keyword synonym clusters, and produce a first cut of a compliance matrix from a NOFO PDF. Then a qualified human verifies everything before it informs a decision.
You will see realistic workflows for each of these tasks, along with the failure modes that AI introduces: hallucinated funder names, fabricated grant amounts, outdated leadership lists, and confident misreadings of NOFO eligibility language. The Human-in-the-Loop principle from Week 1 applies in full: every AI output is reviewed and owned by a person, and the person is accountable for the final research file.
By the end you should be able to use AI to compress a four-hour funder profile into ninety minutes, while running a verification pass that catches the three or four hallucinations every long output produces. The mistake to avoid is trusting AI on facts (amounts, names, dates, eligibility). Trust AI on structure and language, then verify the facts against the 990, the NOFO, and the funder's own published materials.
Common mistakes
These are the traps learners hit most often on this topic. Knowing them in advance is half the fix.
Trusting AI on facts.
AI produces confident, fluent text that is wrong. Names, amounts, dates, and eligibility language must be verified against primary sources every time.
Skipping the verification pass to save time.
AI compresses the first draft, but the verification pass is what makes the output usable. Skipping it converts a fast workflow into a fast way to make expensive mistakes.
Practice problems
Try each on paper first. Click Show solution only after you've made a real attempt.
- Problem 1Describe a two-step AI workflow for producing a verified funder profile, including the verification pass.
Show solution
Step one, paste the foundation's 990-PF and website text into an AI tool and prompt it to produce a one-page funder profile covering capacity, alignment, geography, median gift size, and decision-makers. Step two, run a verification pass that opens the 990 and confirms every dollar amount, every named officer, and the stated charitable purpose verbatim, then opens the foundation website and confirms current contact information and any updated priority language. Any discrepancy is resolved in favor of the source documents, and the profile is signed by the researcher with a date so accountability is recorded.
Practice quiz
- Question 1Which of the following is a SAFE use of AI in prospect research?
- Question 2What is the Human-in-the-Loop principle as applied in this lesson?
- Reflection 3Name three failure modes that AI commonly introduces into prospect research and explain how you verify against each.
Lesson 22 recap
AI accelerates prospect research from hours into minutes when paired with a disciplined Human-in-the-Loop verification pass. Trust AI on structure and language, then verify the facts against the 990, the NOFO, and the funder's own materials.
Coming next: Lesson 23 — Problem Vs. Need
Week 3 turns from prospect research to organizational positioning, where you learn to translate your mission, theory of change, and capacity into the language funders evaluate.
Saved in your browser only — no account, no server.