Paul Strategies X

AI • Tech • Capital • Politics • Fundraising
Christian. Builder. Investor in ideas worth believing

Newsletter 📩 →

[Paul Strategies X @PaulStrategiesX]

Good fundraising copy can still leave the donors cold.

The problem? Most fundraising teams can't see what the donor sees.

And we are building something that can.

[Paul Strategies X @PaulStrategiesX]

Your donor doesn't need convincing.

They already said yes once.

They need reminding why.

[Paul Strategies X @PaulStrategiesX]

https://t.co/Pi4HPil02F
https://t.co/cyMt59Ff2x

[Paul Strategies X @PaulStrategiesX]

Every year-end campaign has someone asking for a bigger match.

Economists Dean Karlan and John List tested it with more than 50,000 prior donors to one nonprofit.

Announcing a match raised revenue per letter by 19% and made people 22% more likely to give.

Making the match bigger, 2:1 or 3:1 instead of 1:1, added nothing.

The comfortable conclusion: donors are doing the math.

My read: they're not. They're reading a signal. Someone else believes in this enough to put money behind it.

This isn't a pricing problem. It's a trust problem.

Here's the practical test: if you have a match this season, spend fewer words on the ratio and more on who the matching donor is and why they said yes.

[Paul Strategies X @PaulStrategiesX]

Google just launched the Gemini Decisions API, powered by a new model called Gemini Decision 1.

Most AI models write answers word by word. This one doesn't write at all. You hand it text, images, audio, video, or a PDF and ask closed questions: yes/no, pick one option (up to 255), or score it on a rubric.

Each answer comes back with a probability and a confidence score, and it can handle up to 1,000 questions about the same input in a single pass.

Pricing: $0.10 per 1M input tokens and $0 for output, which Google says is about 5x cheaper than Flash-Lite. There's also a free tier.

Why it matters: a lot of real work isn't writing. It's sorting. Is this message urgent? Which team should handle it? Does this application meet the criteria?

For a small nonprofit, that could mean triaging volunteer sign-ups or flagging grant reports that are missing pieces, fast and cheap.

Google's own docs say to keep a human on high-stakes calls. I agree. Good stewardship means the tool sorts and people still decide.

Where would a fast yes/no sorter save your team the most time?

[Paul Strategies X @PaulStrategiesX]

@Codie_Sanchez There's real freedom in admitting how much we depend on each other.

Gratitude starts the moment we stop pretending we did any of it alone.

[Paul Strategies X @PaulStrategiesX]

They say donors have short attention spans.

Yet people binge whole seasons about strangers they've never met.

Donors don't have an attention problem.

Your letter has a story problem.

[Paul Strategies X @PaulStrategiesX]

https://t.co/mk26mIz3hW

[Paul Strategies X @PaulStrategiesX]

"I thank my God in all my remembrance of you … because of your partnership in the gospel from the first day until now." Philippians 1:3, 5 (ESV)

Paul didn't thank the Philippians for a transaction. He thanked them for a partnership with a history: from the first day until now.

Most donor communication only remembers the last gift. The kind that builds trust remembers the first day.

Something worth trying: find your longest-standing donor and tell them what has changed since the year they started.

[Paul Strategies X @PaulStrategiesX]

@Codie_Sanchez The third one quietly compounds.

Keeping your word is usually how you earn a seat in the room with the people 10x ahead.

[Paul Strategies X @PaulStrategiesX]

Google DeepMind just announced AlphaProof Code, and it matters for anyone who sends a private message.

The AI system mathematically proved key security properties of Signal's encryption code. Not just the design on paper, but the actual compiled code that runs on devices.

The key details:

• Gemini-powered agents wrote about 6.5 million lines of proof, and every line is machine-checked
• Human experts set the goals and reviewed the specs; agents flagged about 1% of the work for a human second look
• It took four weeks, work that has historically taken expert teams years
• When researchers slipped two old, already-fixed bugs back in, it caught both

It didn't verify the whole Signal app, and it found no new flaws in the parts it checked.

Why it matters: many security bugs hide in the gap between what software should do and what the code actually does. If AI can close that gap cheaply, "proven secure" could become normal.

For nonprofits holding donor records or counseling notes, guarding the people who trust us with their stories is part of faithful stewardship.

Would proof-level security change which tools you trust?

[Paul Strategies X @PaulStrategiesX]

Same paragraph, two audiences.

Active donor: “Because you gave in March, 40 kids started the school year with a reading tutor. I wanted you to know how it’s going.”

Lapsed donor: “It’s been a while since your last gift, and I want to thank you again for it. Here’s what’s happened since.”

One continues a conversation. The other reopens one without guilt.

Gratitude, offered honestly, is how stewardship begins.

[Paul Strategies X @PaulStrategiesX]

https://t.co/XGgJ8OMZBQ · https://t.co/BGxOMMEBIs

[Paul Strategies X @PaulStrategiesX]

A Fidelity Charitable study with VolunteerMatch found that people who had volunteered in the past year gave an average of $2,593 to charity. People who had never volunteered: $230.

Two-thirds of those volunteers said they generally give to the same organizations where they volunteer.

It's an older study (2009). The principle hasn't aged: people give where they've been.

We spend a lot of energy describing the work to donors. Much less letting them stand in it.

Something worth trying: before year-end, invite five donors to see one ordinary hour of your work. Not a gala. An ordinary morning.

[Paul Strategies X @PaulStrategiesX]

@bshlgrs Whatever the odds, the nonprofits I build for need AI that stays accountable to people right now, not someday.

What's the one safeguard you'd want every small organization to insist on today?

[Paul Strategies X @PaulStrategiesX]

@Codie_Sanchez Understanding how someone processes a hard day is half of loving them well.

Listening first usually beats fixing first.

[Paul Strategies X @PaulStrategiesX]

Source: https://t.co/DqA0oY7rrY

[Paul Strategies X @PaulStrategiesX]

In 2000, about two-thirds of U.S. households gave to charity.

In 2020, 46.9% did.

That's from Indiana University's Lilly Family School of Philanthropy. Their 2024 report called it "dollars up, donors down": fewer households giving, but the ones who gave gave more, an average of $3,116, up from $2,792 before the pandemic.

We usually read that as a fundraising problem. It's also a reminder of who a donor is: someone who kept the habit of giving while fewer of their neighbors did.

That's not a name on a list. That's a rare person.

Something worth trying: this week, write to one long-time donor with no ask at all. Tell them what their faithfulness made possible.

[Paul Strategies X @PaulStrategiesX]

A rule I'm holding myself to while building The Coach:

Nothing gets built because it sounds impressive in a demo.

It gets built if a development director at a small nonprofit would actually reach for it on a busy Tuesday afternoon.

Impressive is easy to build. Useful takes listening.

[Paul Strategies X @PaulStrategiesX]

Three words to cut from your next appeal, and what to write instead:

(1) "Beneficiaries." Write the person's name, or what they'd call themselves.
(2) "Capacity." Write what you can do now that you couldn't before.
(3) "Impact." Write what changed, for whom, and when.

Jargon protects the writer. Plain words serve the reader.