What arrived
Any news before the board meets?
A client asks for a status update ahead of a board meeting, landing in the account Beemy already reads.
Beemy · beemy.co
Printed from beemy.co/trust/
Trust & Security
Beemy asks for access to the most private parts of your life. Here's exactly how that access is protected, separated, and kept in your hands.
On this page
The autonomy ladder
Connecting Gmail, your calendar and your docs doesn't hand over control. Beemy starts read-only: watching how you work before it ever acts on your behalf.
Drafts land in your drafts folder. Until you say otherwise, nothing leaves your account without a one-tap review from you.
When you're ready, you decide (category by category) what Beemy can do unsupervised. Everything else keeps waiting for your say-so. Every escalation and every action still carries a plain "why this reached you," and anything Beemy handled on its own stays reviewable after the fact.
One item, shown
Illustrative example, built with mock data.
In Gmail · work
What arrived
Any news before the board meets?
A client asks for a status update ahead of a board meeting, landing in the account Beemy already reads.
Beemy's draft
Re: Any news before the board meets?
A reply is ready, built from this week's own notes, and sits in the drafts folder instead of the client's inbox.
You approve it yourself, whenever you're ready. No deadline sends it for you.
The commitments
Every byte moving between you and Beemy travels over TLS 1.3.
Stored data is encrypted at rest with AES-256, the same standard banks use.
Personal, work and public knowledge live in strictly separated stores. There's no toggle to get wrong.
Your data is never sold, and never used to train shared models.
Ask us to delete your account and your data, drafts and history all go with it. The audit trail proving it happened is the one thing that stays.
Beemy starts read-only. Nothing sends, posts or replies on your behalf until you approve it.
See the exact list of what Beemy will and won't send alone on On Your Behalf.
Who else sees your mail
Some triage and drafting calls run on models that OpenAI and Anthropic operate. Each one sees only the message content of that one call, and nothing else. It keeps that content for no longer than the call takes.
Runs some of the model calls behind triage and drafting, under a zero-data-retention agreement.
Runs some of the model calls behind triage and drafting, under a separate zero-retention agreement.
Both operate under zero data retention and no-training terms, set out in each one's own data-processing agreement, linked above. Neither trains its models on your mail.
Beemy also keeps its own allowlist of endpoints cleared for zero retention. It checks the endpoint on every call, and it refuses any endpoint that is not on the list.
See every vendor, not just these two, on the sub-processors register.
The differentiator
Personal and work knowledge don't share a database with a permission flag on top. They live in physically separate stores. A work reply can never surface your personal life, and a personal message can never surface your work, because there's no path between them to leak across. It isn't a promise Beemy keeps. It's a wall it can't cross.
Curious why we built it this way? Read the beliefs behind this →
The other side of the ledger
And when it does get something wrong within those limits: here's exactly what that costs.
What's connected, and how
Beemy connects to the tools you already use, and reads them read-only at first, before it ever drafts, sends or files anything on your behalf. You can see exactly what's connected. To take one source away, you erase the silo that holds it: see Leaving.
Security FAQ
No. You get a quiet, triaged inbox within your first few minutes connected, and a first draft in your voice to approve that same session. Auto-send comes later, and only for the categories you've promoted on your own autonomy dial. The early wins are fast because they're read-only or approve-first; the trust ladder above still applies underneath them.
No. Beemy never uses your data to train shared models. What it learns is used only to serve you, and stays inside your own silo.
Not until you explicitly allow it, and even then only for the categories you choose (say, meeting confirmations). Everything else waits in your drafts folder for a one-tap review.
Personal, work and public knowledge live in strictly separated stores; there's no shared memory or privacy toggle in the mix. A work reply can't surface your personal life, or vice versa, because there's no path between them.
Your data, drafts and history all go with it. One record stays on purpose: the audit trail that proves the deletion happened. See Leaving for the full exit path.
In transit, everything is encrypted with TLS 1.3. At rest, it's encrypted with AES-256 (the same standard used by banks).
Yes. Every escalation comes with a plain-language "why this reached you," built from the real signals specific to that case. And anything Beemy handled automatically stays reviewable after the fact, so autonomy never becomes a black box.
Autonomy ladder, escalation, silo: the Lexicon defines the trust vocabulary on this page in plain English.
Trust also means the site works for you: what we test for accessibility, and what's still short, on the Accessibility page.
Read next
Beemy is in private beta. Join the waitlist and go from connect to a quiet, triaged inbox before you close your laptop tonight.
You're in the hive. We'll buzz you when it's your turn.
While you waitLinks do not work on paper. Open the address above to follow them.
To join the waitlist, go to beemy.co/waitlist.