The difference between output and correspondence.
Patron researches the prospect, frames the strategy, and drafts in CASL's voice. Then it shows you the letter first, with the brief it was built from and a blunt check of anything that would stop it going out: placeholder junk, a missing citation, an unverified CASL fact, a machine tell like a stray em dash. The diagnostics sit to the side, advisory. Built on codified fundraising practice, not generic AI.
Patron researches and drafts behind the scenes. What it shows you, in order: the letter, the brief it was built from, and a short list of anything that blocks sending.
What the letter was built from. The dossier with its sourced findings and any source conflicts, the strategy, and the keystone: the one fact true of this donor and no other. You read the letter against it.
In CASL's voice, grounded in CASL's programs, with every donor claim cited back to the dossier. Before it is final, a discovery check asks whether one line would land as unexpected and true at once. If not, Patron writes it again.
A blunt, deterministic read of what actually stops a letter going out: placeholder text in the body, a missing or broken citation, an unverified CASL fact, a source conflict, an ask that does not square with the campaign, a machine tell. One status: Hold, worth a look, or clear to send.
Advisory, and tucked in a drawer. The nine craft passes, a donor simulation that reads the letter back as the prospect, and the vital-signs charts. There to consult, never the product.
A draft is easy. The hard part is the seam a real fundraiser spots in two seconds. Patron runs a deterministic check rather than tone scoring, on exactly those.
No template scaffolding left in the body, no unfilled fields slipping through, and no sources stripped or pointing nowhere. If an edit drops a citation, the check says so.
A check enforced in code, so the model cannot wave it off. If the letter states a CASL fact outside the verified set, an unconfirmed wing, a superlative, a reach figure, it holds the letter until a person confirms or removes it.
Em dashes, the antithesis button, the stock flourishes that read as generated. A regex scan, so it cannot be talked out of a flag. A stray em dash holds the letter on its own.
An unresolved source conflict, or an ask that is larger than the campaign or a sliver of it. The numbers and the sources have to agree before it goes.
Most AI tools for fundraising train on what already exists on the open web. Patron trains on the practitioners who codified the craft.
A major-gift letter is not a marketing campaign. It is the closest professional analogue to a private letter from one trustee to another. The donor is the protagonist. The organization is the cause she enabled. The ask is specific, named, and never buried.
The craft library and the send-check are the parts most AI tools skip. A draft is easy. Reading it back as the donor, naming what is unverified, and refusing to let placeholder junk or a machine tell reach the envelope is the difference between output and correspondence.
Every draft uses the names CASL uses internally, never generic nonprofit copy.
Building on the wisdom of generations, CASL catalyzes the transformation of individuals, families, and the community for an equitable future. CASL Mission
Brain & behavioral. Family services. Preventative and coordinated care. The programs a donor like Margaret Yip cares about.
Tutoring, ESL, workforce, mentorship. The programs that follow a CASL family across decades.
The frontline services for new arrivals and elders. The expanded intake space the Hub will house.
The Anti-Hate Action Center. Policy work, research, and the public face of CASL's mission.
Pick a preset donor to see a full run end to end, or enter a real prospect from CASL's portfolio. Drafts stay private; only the people on the Cloudflare Access list can reach the studio.