Skip to content

Guide

How we run five businesses with an AI operations team

· MaxiMind

We run five live products with one AI operations team: AI staff do the drafting and the repeat work, real people check it, and the owner approves anything about money, customers or the law. Every request becomes a numbered task on one task board, so nothing depends on someone remembering it. It is the same team and the same tools we offer customers.

If you run several small brands, the hard part is rarely one big job. It is many small ones, spread across different inboxes, that all need an answer this week. Here is how we keep them in one place, and what you can copy.

Which five businesses do we run this way?

All five are live today, and all five are built and run by the same team and tools:

  • SVFF Online: one-on-one Southern Vietnamese lessons with native teachers.
  • LingoRoom: teaching rooms for language programs, with live lessons, recordings and lesson recaps in one place.
  • Vietlingo: learn Vietnamese by dialect, on the web and on iPhone.
  • Ledoly: English speaking practice: the Sound American podcast and Shadowing+ are live; video courses are coming soon.
  • Ledobit: lessons as adventures for young learners.

They are different businesses with different customers. Behind them, the way the work moves is the same.

How does one owner keep track of five businesses?

Every request becomes a numbered task on one task board. We use MaxiMind Workplace for this. A question from support, a website fix, a new course page, a form to file: each one gets a number, and each task has four things:

  • An owner: the person or AI conversation responsible for it.
  • A status: where the task stands right now.
  • A result with proof: not "done", but what was done and how we know it works, such as a link, a screenshot or a test.
  • A decision: who approved it, or what still needs approving.

The habit behind the board matters: if it isn't on the board, it isn't being worked on, and if it has no proof, it isn't finished.

How is the work split between AI staff?

By job, not by one big assistant. Each business and each department has its own AI conversation: support, website, app, legal paperwork, content. Each conversation has its own written rules.

Why split it up? Because the rules are different. Support replies need to match the tone of one brand. Legal paperwork needs the owner's eyes on every page. A website change needs a check before it goes live. One conversation that tries to do everything mixes those rules up, and mixes up the brands too.

Written rules also mean the work doesn't depend on one person's memory. When a rule changes, we update it in writing, and every new task follows it.

Who checks the AI's work?

Every task passes through three layers:

  • AI drafts. Replies, pages, posts, reports, paperwork: the first version comes from the AI staff.
  • A person checks. Someone reads the draft and the proof before it moves on.
  • The owner approves. Anything about money, customers or the law waits for the owner's yes.

Customer messages stay drafts until a person sends them. The AI never messages a customer on its own.

What happens when the AI gets something wrong?

It does get things wrong sometimes, so we don't rely on it being right. We rely on guards that stop a mistake from spreading:

  • Only the owner is admin. No AI conversation and no staff member can give itself more access.
  • Each person sees only their own tasks and files. Work for one brand doesn't leak into another, and nobody sees more than their job needs.
  • Spending needs the owner's yes. No AI conversation and no staff member spends money without it.
  • Proof before "done". A result without proof goes back for another round.

When a mistake does get through, fixing it becomes a task like any other, and the lesson goes into the written rules.

What can you copy for your own brands?

You don't need five businesses to use this. The same steps work for two brands, or for one shop with several jobs:

  • Step 1. Put every request on one board, with a number. Messages, emails, notes on paper: if it needs doing, it becomes a task.
  • Step 2. Give every task an owner, a status, a result and a decision. Then you can see what is stuck without asking anyone.
  • Step 3. Split the work by job. Support, content, bookings, paperwork. Write a short set of rules for each.
  • Step 4. Decide what always waits for you. Money, customers and the law are a good start.
  • Step 5. Ask for proof, not "done". A link, a screenshot, the actual message that went out.

Is this the same AI operations team MaxiMind offers?

Yes. The team, the task board and the rules above are what we set up for customers: an AI team lead and AI staff for each job, supervised by real people, with the owner approving decisions on their phone. Today the AI team works in the appointment book and on the task board. Website support chat and phone answering are coming soon.

See the plans at MaxiMind's AI operations team. Send us a few lines at maximind.dev.


Questions

Does AI run your businesses on its own?

No. AI drafts and does repeat work. A person checks every result, and the owner approves anything about money, customers or the law.

What is MaxiMind Workplace?

It is the task board we use for all five products. Every request becomes a numbered task with an owner, a status, a result with proof and a decision.

Does the AI message your customers directly?

No. Customer messages are drafts until a person reads them and sends them.

Why use separate AI conversations instead of one assistant?

Because each job has different rules. Support, website, app, legal paperwork and content each get their own conversation with their own written rules, so the rules and the brands don't get mixed up.

Do I need several brands to use an AI operations team?

No. The same habits work for a single business. On the AI operations page, the Scale plan covers several locations; the other plans suit a single business.

Tell us about your business

Send a few lines and we will read them and get back to you to set up a call.

Online booking opens soon. Until then, the booking button opens our contact form and we agree a time with you.

Đọc bằng tiếng Việt