Privacy
What we read, and what we never do with it.
You are handing us the inside of your company's mail. This page says plainly what happens to it. No defined terms, no schedules.Last updated July 2026
Read-only, always.
The connection we ask for cannot send, delete, or edit anything. We never send mail from your account.Nobody here reads your email.
Software reads it and returns counts and patterns. A human at Lemonbrand only sees the map, and only what you show us.Your data never trains a model.
Not ours, not a vendor’s. Training is disabled on every model connection we run.You can revoke it in one click.
From your own Google or Microsoft admin, without asking us. The read stops immediately.- What does the free sample actually read?
- The last three weeks of one inbox you choose: headers, subjects, timestamps, participants, and message bodies — needed to tell a repeated request from a one-off. Attachments are not opened.
- And the paid audit?
- Six months across the inboxes your team connects. Same scope per inbox, wider view, because handoffs only show up between people.
- What do you keep afterwards?
- The derived map: clusters, counts, and the workflows we recommend. Raw message content is discarded after the audit is delivered and is never used for anything else.
- How long is any of it stored?
- Derived audit data for 12 months, so we can compare before and after. Ask us to delete it sooner and we do, within seven days.
- Do you sell or share it?
- No. There is no advertising, no data brokerage, and no analytics vendor with access to message content.
- Who owns the agents we build?
- You do. They run in your own Claude or ChatGPT workspace, on your accounts. If you stop working with us, they keep running.
- Where is it processed?
- In North American regions of our vendors. Tell us you need a specific region and we will say yes or no before you connect anything.
- What if there is a breach?
- We tell you within 72 hours of finding out, with what was in scope and what we did. In plain language, from a person.
Who else touches it
Three vendors, and what each one is for.
- Pipedream
- Holds the read-only connection to your mail provider and moves messages to the audit.SOC 2
- Anthropic
- Claude reads and clusters the threads, and runs the agents we build with you.No training on your data
- OpenAI
- Used where a workflow you asked for lives in ChatGPT or Codex instead.No training on your data
Still want it in writing?
Ask and we send the data processing agreement and the deletion form. Both are one page.Contact usWhat we read, and what we never do with it.
You are handing us the inside of your company's mail. This page says plainly what happens to it. No defined terms, no schedules.Last updated July 2026Read-only, always.
The connection we ask for cannot send, delete, or edit anything. We never send mail from your account.Nobody here reads your email.
Software reads it and returns counts and patterns. A human at Lemonbrand only sees the map, and only what you show us.Your data never trains a model.
Not ours, not a vendor’s. Training is disabled on every model connection we run.You can revoke it in one click.
From your own Google or Microsoft admin, without asking us. The read stops immediately.- What does the free sample actually read?
- The last three weeks of one inbox you choose: headers, subjects, timestamps, participants, and message bodies — needed to tell a repeated request from a one-off. Attachments are not opened.
- And the paid audit?
- Six months across the inboxes your team connects. Same scope per inbox, wider view, because handoffs only show up between people.
- What do you keep afterwards?
- The derived map: clusters, counts, and the workflows we recommend. Raw message content is discarded after the audit is delivered and is never used for anything else.
- How long is any of it stored?
- Derived audit data for 12 months, so we can compare before and after. Ask us to delete it sooner and we do, within seven days.
- Do you sell or share it?
- No. There is no advertising, no data brokerage, and no analytics vendor with access to message content.
- Who owns the agents we build?
- You do. They run in your own Claude or ChatGPT workspace, on your accounts. If you stop working with us, they keep running.
- Where is it processed?
- In North American regions of our vendors. Tell us you need a specific region and we will say yes or no before you connect anything.
- What if there is a breach?
- We tell you within 72 hours of finding out, with what was in scope and what we did. In plain language, from a person.
Who else touches it
Three vendors, and what each one is for.
- Pipedream
- Holds the read-only connection to your mail provider and moves messages to the audit.SOC 2
- Anthropic
- Claude reads and clusters the threads, and runs the agents we build with you.No training on your data
- OpenAI
- Used where a workflow you asked for lives in ChatGPT or Codex instead.No training on your data
YouTube
LinkedIn
TikTok
Instagram
Facebook
Substack