Skip to content
CasePrompts

How your data is handled

Your data stays on your side.

CasePrompts is built local-first: the work happens in your browser, a server is used only when you choose, and only through providers held to zero-data-retention agreements. Here's exactly what that means — and what it looks like.

01

The builder runs in your browser

Everything you type into the builder — party names, issues, the facts of the matter — is assembled into a finished prompt right on your own device. There is no upload, and for the core tool there is no server in the loop at all. Close the tab and it's gone; the only thing that persists is what your browser remembers for you, like your last-used practice.

caseprompts.com — your browser

Your inputs

names, issues, facts

CITE assembly

on your device

Finished prompt

ready to copy

Nothing here is uploaded — it never leaves your device.

A browser window showing your inputs flowing into CITE assembly and a finished prompt, all inside your own device, with a note that nothing is uploaded.

02

A server is used only when you choose

The free builder and the whole library never call a server. The only features that do are the optional AI tools — CITE Coach and Prompt Studio — and they run only when you click the button. No background calls, no automatic sending. The decision to involve a server is always an explicit one you make.

Your browser

the prompt you built

You click “Run”

the gate — nothing passes without it

AI provider

processes the request

Skip the AI tools and the path simply never opens. The free builder lives entirely in the first box.

A left-to-right flow: your browser, then a gate labelled "you click Run — nothing passes without it", then the AI provider. Without your click, nothing flows past the gate.

03

The provider keeps nothing

When you do run an AI feature, the prompt text goes to an AI provider under a zero-data-retention agreement: they process the request, return the result, and retain nothing afterward. We don't keep a copy on our side either, and your work is never used to train a model. There's still no document upload anywhere — the tools work on the prompt text itself, not your files.

Your prompt

sent on your click

Provider processes it

generates a response

Result back to you

in your browser

Retained afterward: nothing. No stored prompts, no copy of your record, no training on your data.

A flow showing your prompt sent to a provider, processed, and the result returned, with a callout that nothing is retained afterward.

What we never do

  • Upload or store your case files — there is no upload feature at all
  • Keep a copy of the prompts you build
  • Send anything to a server for the free builder
  • Use your work to train models

Built so you don't have to choose between useful and careful.

The principle is simple: keep the work on your device, make any server call a deliberate choice, and partner only with providers who keep nothing. You still own the duty to verify every output against the record — these choices just make sure the record stays yours while you do.