Where AI shows up. And where it stops.
Xerish uses AI in a handful of narrow, clearly bounded places — as a stewardship tool, never an authority. Here is every surface it appears on, what powers it, what it can see, and the lines it will never cross.
Every surface, named.
There is no hidden AI on Xerish. It appears in exactly these places, each a narrow tool with its own guardrails.
Treasury insights. Short stewardship reflections in your Treasury, drawn from your own giving history.
The giving coach in the app. A bounded assistant for signed-in Givers that answers questions about your own giving.
Growth coaching for organizations, and formation coaching for churches. Data-grounded advisors that help an organization or congregation understand its own activity on Xerish — never anyone else’s.
An internal analytics assistant. Used by the Xerish team to read platform-level trends. Internal only — it never speaks to Givers.
The coach on this website. A public help assistant that answers questions about how Xerish works. It requires no account and sees no account data.
None of these is a chatbot companion. Each surfaces information and stops. If AI ever appears somewhere new, this page will name it.
Anthropic Claude, and nothing homegrown.
Every AI surface on Xerish is powered by Claude, the AI model built by Anthropic, reached through Anthropic’s API. Xerish does not build or operate its own AI models. When you ask the coach a question, Xerish assembles a small, purpose-built context, sends it to the Claude API, and shows you the answer.
Only what the answer needs. Only yours.
Each request carries the minimum context needed to answer — nothing more. For a signed-in Giver, that means aggregates from your own account: your giving pace and totals, a goal you’ve set, your active recurring gifts, the causes you support, and the impact tallies you already see on your own screens. It is the same data your Treasury shows you.
It never sees other Givers’ data. The context is built from your account alone — there is no pool of everyone’s giving for it to draw comparisons from, by design. Organization and church coaches work the same way, on that organization’s own activity. And the website coach is fully public: it answers from Xerish’s published materials and sees no account data at all.
What the AI is forbidden to do.
These boundaries are written into every AI surface as standing instructions — part of the product, not a promise bolted on afterward. The Xerish AI will never:
- Tell you how much to give. That is between you and God.
- Compare you to other Givers, or rank you against anyone.
- Manufacture urgency or guilt to push you toward giving more.
- Interpret Scripture authoritatively, or stand in for a pastor or counselor.
- Give tax advice. For that, talk to your own advisor.
- Make grant decisions. The AI holds no authority over money — every grant flows through Xerish Storehouse’s approval framework, with human oversight. See who controls the money.
If you ever see the AI cross one of these lines, that is a bug — tell us at contact@xerish.com and we’ll fix it.
Processed to answer you. Not to build a model.
Your questions and the context behind them are processed through Anthropic’s Claude API to generate the answer, under the confidentiality terms described in the Privacy Policy. Xerish does not train AI models on your data. There is no Xerish model, and your giving history is not used to build one. Your data’s job is to answer your question — then it goes back to being yours.
The wider picture — what Xerish collects, where it lives, and the controls you hold — is on the Your Data & AI page.
Reach a human.
Questions about AI on Xerish — how it works, what it saw, or anything this page didn’t answer — email contact@xerish.com and a person will answer.