You decide how far it goes
We build each agent around the oversight you want, from approving every change to letting it run on its own inside limits you set. Every move is logged and reported with the numbers behind it.
Three levels of oversight
You choose the level when we scope the agent. Today the agent runs on Shopify App Store ads with approvals on.
Approve every change
Every change waits for your yes. Nothing moves without you, and a yes covers that change only.
Act inside limits
It makes the changes your limits allow, and logs each one. Anything outside a limit waits for you.
Run on its own
It makes every change itself, inside the limits and permissions you set, and reports each move with the numbers behind it.
What holds whatever you choose
It can't go past a limit
The agent can't go past a limit, or use a permission you blocked, at any level.
Permissions are yours
Budgets, billing, credits, creating or archiving ads and resuming paused ads are yours to allow or block.
Every change is logged
Old value, new value, reason and rule, so it is clear what to put back.
The evidence is shown
Each change names the rule, the numbers and a confidence level. Thin data is flagged as an early signal and is not acted on.
How a change moves from idea to log
- [01]
The agent reviews
It reads the numbers and checks them against your written rulebook.
- [02]
It decides
It checks each change against your limits and permissions. Inside them it goes ahead, unless approvals are on. Outside them it asks you.
- [03]
You answer when asked
You approve or decline. A yes covers that change only and does not carry to the next review.
- [04]
It applies and verifies
The change is applied, then the value is re-read to confirm it.
- [05]
It logs and reports
The log keeps old value, new value, reason, rule, expected result and dates to check the outcome. The report gives you the numbers.
Every change is written down
The change log records what changed, from what to what, why, under which rule, what we expect, and when to check. The dashboard shows a brief, signals, changes, tests and learnings, and is updated after each review. You can ask it questions in plain language, and the chat is read-only.
A Heyo Growth Engineer supervises every agent
Heyo agents are scoped, built and supervised by Heyo, at whatever level of oversight you choose. A Heyo Growth Engineer watches how each one runs and is who you talk to. The limits, the permissions and who approves on your side are agreed in the scope.
Questions people ask first
Can an agent run up my spend?
It can't go past the limits you set, and it can't touch budgets unless you allow it.
Is it a black box?
No. Every change states the rule, the evidence and a confidence level, and every change is logged.
Will I lose control of my accounts?
No. The agent can't go past the limits you set, every change is logged, and a Heyo Growth Engineer supervises it. Choose approve every change and nothing moves without you. If you want it to stop, tell us.
Could it put my account at risk?
Any tool that changes an account carries some risk, and each platform has its own rules. We agree the limits and permissions with you in the scope. We cannot promise how a platform will treat an account.
What happens if it gets something wrong?
Every change is logged with the old value, so it is clear what to put back, and the limits cap how far a mistake can go. With approvals on, you can also decline a change before it happens.
Does it run on its own?
If you want it to. We build it to the oversight you choose: approve every change, let it act inside limits, or let it run on its own. A Heyo Growth Engineer supervises it either way.
Can I see one running?
Today it runs on Shopify App Store ads with approvals on. Other setups are scoped and built for your account.
Who supervises it?
A Heyo Growth Engineer scopes, builds and supervises every agent, whatever level of oversight you choose.
Can you show results?
It is early. What we can show today is how it works and an example review.
Talk through your controls
Book a call to agree the oversight level, the limits and the permissions, and how every change is logged.