The principle. A worship song is the labor of the composer, the
publisher, and the congregation that has carried it. A sermon is the labor of the
preacher and the local church that received it. Those words belong to them — not
to us. It would not be honoring to the creative work of composers or to the
pastoral work of preachers for Reed Light Books to warehouse their material on
our servers, much less to study or train against it.
It is also not in our best interest to hold that material. A server full
of pastoral writing is both a security target and a temptation. The cleanest way
to behave well is to never possess it in the first place — analyze the text,
return the result, let the words go.
What never touches our disks.
The following is true for every analysis run by every user, on every tool in the
apps suite. None of it is written to a database, a log file, a backup, or a cache:
We do NOT store
- The full text of any worship song you paste, search for, or fetch by URL.
- The full text of any sermon, outline, notes, or imagery fragment you submit to Sermon Lens.
- Snippets, excerpts, or quoted phrases drawn from your sermons or lyrics for any later use.
- URLs you fetch through the Search-by-URL or Search-by-Title flows.
- Any congregational note you write into the Sermon Lens picker beyond the duration of the single analysis.
- Any of the above for the purpose of training a model, profiling you, advertising, or sale. Ever.
We DO keep, narrowly
- Your account record — name (optional), email, password hash, coin balance, the last ~50 coin transactions on your line. No content.
- Request logs — IP, timestamp, response code, path, browser user-agent. 30-day retention. No request body.
- Usage logs — per-action records: which tool, coin cost, timestamp, your user ID. No lyric or sermon text.
- Audit logs — sign-in, password reset, admin actions. No content.
- Sermon Lens history (per-user) — the AI's analysis notes from sermons you've run, so you can revisit them in your sidebar. See the per-tool walkthrough below for what is and isn't in there.
Per-tool walkthrough.
Each tool handles the words you give it the same way at the bottom of the stack — send, analyze, return, discard — but the persistence story above that differs in small, honest ways:
About the sample sermons in the picker.
Sermon Lens ships with four pre-loaded sample sermons so a first-time visitor can
try the tool without uploading their own work. Those four are all publicly
available material, used here for demonstration:
- The Sermon on the Mount — Matthew 5–7, from the ESV.
- Sinners in the Hands of an Angry God — Jonathan Edwards, 1741, public domain.
- Don't Be No Fool — Charlie Dates, from a published transcript.
- The Power of the Gospel — Tim Keller, from a published transcript.
These ship as static files inside the app. They are not user data, and using them
does not consume coins. The pre-baked analyses you see when you open a sample are
fixtures we prepared once — the model is not being re-run on them every time.
Third parties that touch your request.
To deliver the analysis, your request necessarily passes through a few services. Each one is listed below with what it sees and what it doesn't:
-
Anthropic. The AI provider that performs the analysis. The text
you submit is included in the API call. We use Anthropic's commercial API, under
which Anthropic does not train models on submitted inputs.
Anthropic's own handling is governed by its
privacy policy and the
commercial API terms.
-
Hymnary.org (via the
api.allorigins.win CORS proxy).
When you use Clear Worship's Search by Title, your browser sends only
the title string. No credentials, no lyrics, no account information.
-
Stripe. Handles coin-pack and subscription purchases. Card data
goes directly from your browser to Stripe and is never visible to our server. We
receive only a session ID and a confirmation event from Stripe's webhooks.
-
Cloudflare Turnstile. A bot-check on sign-up and sign-in. It
checks a token, not your content.
-
SMTP mail provider (
mail.supremecluster.com).
Delivers transactional email — verification, password reset, receipts. Carries
only the message body of those emails, never your lyrics or sermons.
Why we do it this way.
First, because it is honoring. The worship songs we analyze
are the labor of the people who wrote them, the publishers who carry them, and
the congregations that have sung them into being. The sermons we analyze are the
labor of preachers and the local churches that received them. Those words are
theirs. We are a tool that helps you think about them — not a vault that owns
them.
Second, because it is wise. A growing pile of pastoral writing
on our servers would be both a security target and a temptation. The cleanest
way to refuse temptation is to never possess the material. So we don't.
Third, because it makes our promises easy to keep.
"We don't sell your data" is a much harder promise to keep when you have a lot
of it sitting around. The discipline of not storing it in the first place keeps
us honest by default.
Changes to this practice.
If we ever change how data is handled — add a feature that requires storing
something new, change a retention window, swap a third-party service — we will
update this page and note the change in
the public changelog. The effective date at the top
will reflect the most recent revision. Material changes will also be flagged on
your next sign-in.