Agent setup
A written charter for each agent: what it owns, what good output looks like, and a clear list of things it must never do. One role per agent, with an optional manager agent that routes work to the right place.
AI agent setup · Any platform
AI agents that work for you, and wait for your approval.
I set up AI agents for founders, solo operators and small teams on the platforms and tools you already use. Each agent gets a clear role, routines that run on time, and approval gates, so nothing is sent, posted, spent or deleted without your yes.
0
Sends without
your yes
your yes
1
Phrase stops
every task
every task
Any
Agent platform
or toolset
or toolset
The problem
AI agents are powerful. Without guardrails, they create chaos.
Most people try an agent, hand it access, and hope for the best. Then the problems start.
01
It takes actions nobody approved
Agents delete files you needed, "fix" failing tests by removing them, or send messages nobody signed off on.
02
It leaves a mess behind
Clutter piles up and never gets cleaned. One agent juggles five jobs, so when something breaks, you can't tell why.
03
It costs you time instead of saving it
Instead of delegating, you end up supervising every step and double-checking everything it does.
The cause is usually simple: nobody told the agent what it owns, what it must never do, and when it has to stop and ask. That is exactly what I set up.
What you get
A setup that does the work and checks with you first.
Every setup is built around how you actually work, using the agent platform and tools that fit your business.
01
02
Routines that run on time
Choose from a morning brief, a Friday review, an end-of-day handoff, a single reminder for approvals left waiting 24 hours, and a monthly routine audit.
03
Approval gates
Every action is sorted by risk. Reading and drafting happen automatically. Sending, posting, spending and deleting always wait for your approval.
04
House rules and a STOP ALL switch
Up to 10 plain-language rules, such as "If it can't be undone in under a minute, ask first." Agents log their work and show their sources. One phrase halts every task and reports what was in progress.
How approvals work
The two riskiest levels never run without youL0Read and researchAutomaticL1Write drafts and filesAutomaticL2Reversible changes within the agent's assigned jobWithin scopeL3Send, post or change a live accountNeeds your approvalL4Spend money, delete, or anything that can't be undoneNeeds your approval
How it works
Three steps to an agent you can trust.
You stay in control the whole way, and you hand over responsibility gradually.
01
Free fit call
Tell me what slows you down
We talk through your work, the tools and agents you already use, and what you would hand off first. If an agent isn't the right fix, I'll tell you.
02
Setup
I set up your agents
Charters, house rules, approval gates, routines and the STOP ALL switch, configured for your workflow. We test the stop switch together at a live handover.
03
Gradual handover
You hand over control step by step
Over four weeks, your agents move from drafting only to running on schedule. We meet at the two-week mark to fix any rule that misfired.
Week 1The agent only drafts. You review everything.
Week 2You approve each action.
Week 3It handles routine work and flags exceptions.
Week 4It runs on schedule. You read a weekly summary.
Why trust this approach
Built so you stay in charge of your agents.
I use this same approach in my own work. For you, it means each agent has one clear role, routines keep the work on schedule, and you approve everything that goes out.
- One role per agent. When something goes wrong, you know exactly where to look.
- Daily briefings. You get a short summary instead of watching every step.
- Nothing goes out without your approval. Sending, posting, spending and deleting always wait for your yes.
Tiers
Start with one agent or build a full team.
Every setup tier includes a live STOP ALL test and a two-week check-in. We'll find the right fit, and discuss pricing, on your free call.
01
Agent Safety Audit
For an AI agent you already run, on any platform.
- Review of one existing agent
- House rules review
- Approval gates review
- STOP ALL review
02
Solo Guardrails
One agent, set up properly from day one.
- One agent with a charter
- 10 house rules
- Approval gates
- 3 routines
- STOP ALL and a two‑week check‑in
03
Team Guardrails
Several agents working under one set of rules.
- Up to 3 agents or workflows
- Shared house rules
- Approval gates for each agent
- 5 routines
- STOP ALL across every agent and a two‑week check‑in
04Most complete
Manager Bot Build
A manager agent that coordinates a small team of specialists.
- A manager agent and up to 3 specialists
- Routing so work reaches the right agent
- Work logs and source review
- 5 routines and STOP ALL
- Two-week check-in and 30 days of fixes
Not included: custom software development, 24/7 monitoring, or buying tools on your behalf. You keep your own accounts and pay any subscriptions directly.
FAQ
Questions, answered.
Is it safe? Can an agent send something without me?
Not when it is set up properly, and that is the whole point of this service. Reading and drafting happen automatically, but sending, posting, changing live accounts, spending money and deleting always wait for your approval. Every agent logs its work, and a single STOP ALL phrase halts everything and reports what was in progress. You also hand over control gradually over four weeks, starting with drafts only.
Which AI platforms and tools do you work with?
The approach is not tied to one platform. Charters, house rules, approval gates, routines and a stop switch apply to any AI agent. On the fit call, we look at the agent platform you already use, or choose one that suits your work, and decide which of your tools (such as email, social accounts, calendars and schedulers) each agent may read and which it may change. You keep your own accounts and pay any subscriptions directly.
Do you need my passwords?
No. Passwords and codes should never be shared in chat. Wherever a sign-in is needed, you complete it yourself. Each agent gets only the access its job requires, and read-only access wherever reading is enough.
How long does setup take?
It depends on the tier and how quickly we complete your intake questions. Once your agents are set up, the four-week gradual handover begins, with a check-in at the two-week mark.
I'm not technical. Is that a problem?
Not at all. You work with your agents in plain language, and the house rules are written in plain language too. You never need to write code.
What happens on the free fit call?
We talk through what takes up your time, what you would hand to an agent first, and which tier fits, if any. There is no obligation.
Get started
Stop supervising. Start delegating.
Tell me what you would hand off first. I'll give you an honest answer on whether AI agents are the right fix.
Email moprofitz@mail.grokbot.com or find me on X at @ZyMoProfit.