Skip to content

Security & privacy

Your clients' information stays in your systems.

We work on how your firm operates — its knowledge, its process, who can reach what — not on your clients' records. Here is exactly what we touch, what we keep, what we never do, and how this website handles data. It's one page so you can read it, or paste it into your own AI and ask it questions.

Last updated: 2026-10-22. A plain description of how things work today, not a privacy policy.


Your data stays with you

We don't pull records out of your systems, and we don't store your clients' information. What we build from is prepared with you, inside your systems, with everything identifying removed. The assistant answers from that approved copy and has no login to your systems; anything your people's own Claude or ChatGPT can reach is under their account and your controls, not ours.

Never pooled with other firms

Your firm's version of the assistant is reachable only by your people. What we share across firms is general practice we write ourselves — never another firm's content.

Metadata, not conversations

We see which tools were used, when and how well — not what people asked or what they were told. The one exception is below, stated plainly.


What happens, by kind of work

Your firm's knowledge, and the assistant that answers from it

We build it with you, in conversation and from snapshots you prepare with us inside your own systems. There is no direct integration pulling data out.

Everything identifying is removed or made generic before it reaches us — names, phone numbers, email addresses, account numbers, and anything personal, health-related or otherwise sensitive. We keep what makes it realistic for your people: for a brokerage, the real neighborhoods of your city; for any firm, how you actually do things.

The result is the knowledge your assistant answers from. It describes your process; it contains no one's personal information.

Connecting the tools you already have

We help configure the connectors your Claude or ChatGPT workspace already offers — to Microsoft 365 or Google Workspace — using only what your IT team already controls, in their admin consoles and with them.

We scope them to your firm's knowledge and procedures: the documentation of how you work, not your client records. Your CRM and other line-of-business systems are not connected at this stage.

Custom connections to your systems

When a firm wants AI to reach its CRM or other systems, we write the code and your team runs it in your own infrastructure — AWS, Google Cloud, Azure or whatever you use. We may keep a copy of the source code; we never take or store your data.

Because it runs in your environment, its logs and telemetry belong to your security team, not to us. This isn't a plan: it's how our first client's custom work runs today, in their own cloud account.


Exactly what Zero Mission keeps

  • Who is connected — names and work email addresses, the address each invitation went to, and what each person can reach, so your firm's owners and managers can see it and switch anyone off.
  • Usage metadata — which tools were used, when, how often, whether through Claude, ChatGPT or the iPhone app, and which part of your firm's guidance answered — never the question itself. Your firm's owners and managers see it per person, and can pause the assistant for everyone.
  • Who is on which engagement — if your firm uses engagements: the name you give each piece of client work, who is on it and from when to when. Not the work itself; that stays in your systems.
  • Questions the assistant couldn't answer — kept so we can cover them in the next update, and visible to your firm and to us. We ask people to leave client details out of every question.
  • Ratings and feedback — satisfaction ratings, any comment added to them, and feedback people choose to send us.
  • Inquiries — what someone types into the inquiry form on this site, below.

What we don't keep: the questions the assistant answered, or its answers. Those stay between your people and the AI provider, or on their phone.

The AI providers

Conversations happen in Claude or ChatGPT, under your firm's account. We set firms up on business plans whose terms don't use your content to train their models by default, and we'll show you where each provider says so.

Contractors who connect from their own personal account are under that account's terms — one more reason client details stay out of questions.


Accounts, sign-in and the assistant

  • Accounts are created by invitation only. If you sign in with Google, we receive your name and email address from Google to identify your account.
  • Your account belongs to your firm. Its owners and managers can see who is on its list and what each person can reach. Billing and invoices are visible to owners and managers only.
  • When you connect the Zero Mission assistant to Claude or ChatGPT, you authorize it here. We issue an access token scoped to your firm. We never receive your Google or Microsoft password, and the token can be revoked at any time — removing someone from your firm's list revokes it immediately.

The iPhone app

  • You sign in to each firm the same way, by invitation, and the app keeps that sign-in in the iPhone's Keychain.
  • Your firm's guide is downloaded to the phone, so it answers without a connection: a question it can answer is answered on the phone and isn't sent to us. The copy is protected by the phone's own encryption while it's locked, and stops answering seven days after it was last updated. If someone is removed from your firm's list, the copy is deleted the next time the app checks in.
  • Your conversations stay on the phone.
  • When it's online, the app sends us the same things the assistant does elsewhere: a question it couldn't answer goes to your firm's list, with the question, and for one it did answer, which part of the guide answered — never the question.
  • It also sends diagnostic traces, so we can see what's slow or failing: which screens were used, whether a question was answered, counts and timings, the app's version and the phone's model and iOS version. Never the text of a question or an answer — the server discards anything else a trace carries.
  • Talk it through copies the question and your firm's answer and opens the AI assistant you chose. From there, the conversation is under that assistant's account and terms, as any conversation you start there is.
  • Demo firms — a generic guide for a kind of firm, to try the app — work without signing in. Nothing you ask in a demo is sent to us.
  • There is no advertising, and no third-party analytics or tracking, in the app.

This website

  • There is no third-party analytics, advertising or tracking on this site. The only cookies are the one that protects the forms and, if you sign in, the one that keeps you signed in.
  • The site runs on Google Cloud in the United States. Traffic reaches it through Cloudflare, which carries it the way any web host's network does — your address passes through it to reach us.
  • The Try It prompt on our industry pages runs in your own Claude or ChatGPT, under your account there. Nothing from that conversation is sent to us unless you choose to paste its summary into our inquiry form.
  • What you type into the inquiry form — your name, company, phone, email and message — is stored so we can reply and keep track of the conversation, with the page you wrote from.
  • To stop the form being used by automated senders, we store a one-way fingerprint of your network address, with no way to recover the address from it, and count how many submissions it has made in the past hour. Those records are deleted automatically after an hour and are never linked to your inquiry.

What we don't promise. Using Zero Mission doesn't by itself make anything privileged, compliant or confidential, and it doesn't replace your security stack or your professional obligations. We configure carefully, keep it that way, and tell you plainly where the limits are. Questions: [email protected].

Questions about security?

Ask anything about how your firm's information would be handled — before anything else happens. If your security team wants to talk to us directly, that's welcome.