How we keep your data safe

AI agents only help if you can trust them with your business. Here is exactly how we set them up, what they can touch, and what we never do.

Our practices

Six ways you stay in control

Least-access accounts

Each agent gets its own account with only the access its job needs. No shared logins, and no admin rights when a narrower role will do.

Your data stays in your tools

Agents work inside the systems you already run. Your records stay where they are, and you own everything we build.

Every agent action logged

Agents keep a record of what they read, drafted and changed, so you can see what happened and when.

A person approves before anything goes out

Emails to customers, quotes, invoices and changes to your records wait for a person on your team to say yes.

Month-to-month, easy offboarding

No long-term contract. If you leave, we hand everything over and remove our access.

Plain-English answers

Ask how any workflow handles your data and you get a straight answer from the person who built it.

How access works

Set up so you can check it yourself

Everything we set up shows up in your own admin and activity settings, so you don't have to take our word for it.

  1. 01

    Map the workflow

    On the mini-audit and kickoff we list exactly which tools and records a workflow needs, and which it doesn't.

  2. 02

    Create scoped accounts

    Agents get their own accounts with the narrowest roles that work, set up through your tools' user settings.

  3. 03

    Approve and review

    Agents draft, a person approves, and the log shows what each agent did.

  4. 04

    Offboard cleanly

    If you leave, we hand over the workflows and notes and remove every account we created.

What we never do

Lines we don’t cross

  • Sell or share your dataYour data is used to run your workflows and nothing else.
  • Let an agent act on its own where you want a checkIf you've asked for approval on something, it waits for a person.
  • Ask for passwords over email or textAccess is set up through your tools' own user and sharing settings.
  • Use an admin account when a narrower one worksAgents get the smallest role that does the job.
  • Lock you inNo long-term contracts and no proprietary system you can't take with you.
  • Claim certifications we don't holdWe describe what we actually do, and you can check it.

Have a security questionnaire?

Send it over and Brooks will answer it in plain English, including the parts where the honest answer is “we don’t do that.”

FAQ

Do you hold security certifications like SOC 2?

No. Purple AI is a small, founder-led firm and we don't hold security certifications. What we offer instead are the practices on this page, and you can check most of them yourself in your own tools' admin and activity settings.

Where does my data live?

In the tools you already use, like your email, calendar, files and accounting software. Agents read from and write to those systems. We don't move your records into a separate system of ours.

Is my data used to train AI models?

We don't train models on your data. We'll tell you which AI providers each workflow uses, and where a provider offers business settings that keep your data out of model training, we use them.

Who at Purple AI can access my systems?

Only the accounts we set up with you, with the access each workflow needs. You can see those accounts in your own admin settings and remove them at any time.

Can we start with approval on everything?

Yes, and most clients do. Agents draft and a person approves. As you get comfortable you decide what, if anything, can run without a check.

What happens if we cancel?

Plans are month-to-month. We hand over what we built and the notes on how it works, remove our access, and you confirm it's gone in your own settings.

Find your first AI win
in 15 minutes

Book a free AI mini-audit with Brooks and leave with a plan for your first agent.

Or email [email protected] · (952) 201-1349