An agent reads your real data, decides what to do, then waits for your word.
No building, no prompts to write, no code. You answer a short form, and the agent does the job you picked it for, on the schedule you chose.
What happens on every run
The same five steps, whichever agent it is. You can read every one of them afterwards in your activity log.
-
STEP 1
Something starts it
A schedule falls due, an enquiry lands at your forwarding address or your website form, or you press Run now.
-
STEP 2
It gathers the facts
It reads only what you connected: the enquiry itself, your own web pages, your Search Console data, or a Google Sheet you chose.
-
STEP 3
It decides
Your business profile, your tone and your setup answers go to the AI with the data, and it produces a plan: what it found and what it wants to do.
-
STEP 4
It acts, or asks
Reading and reports happen on their own. Anything that sends, posts or changes something waits in your approval queue until you allow it.
-
STEP 5
It writes down what it did
Every step in plain language, with the reason, so nothing an agent does is a mystery.
Three limits that keep agents safe
Set once, enforced on every run, whatever the AI decides.
Approval mode
Every agent starts by asking. You approve, edit or reject each draft. When you trust a task, switch that one to automatic, and nothing else changes.
A daily limit
Each agent has a cap on how many actions it may take in a day. Reach it and the agent stops until tomorrow, even if work is waiting.
Monthly runs
Your plan includes a number of runs. When they are used up, agents pause until next month. You are never billed for more without asking.
What an agent can do
- Read enquiries sent to your forwarding address or website form
- Read your own public web pages
- Read your Google Search Console data, with read-only access
- Read and add rows to a Google Sheet you choose
- Draft and send email, once you allow it
- Plan a follow-up for a later day
What it cannot do
- Use a tool the agent was not given
- Read an account you have not connected
- Change anything on your website
- Send a thing while it is in approval mode
- Act after its daily limit, or after your monthly runs
- Invent a price, a date or a promise that is not in your details
Your data
Only what you connect
An agent sees your business profile, your setup answers and the one account it needs. Disconnect it and the agent pauses.
Sent to Gemini to think
The data an agent reads goes to Google Gemini to produce a reply. It is not used to train models, and it is not sold.
Deleted on your schedule
Activity history is kept for the period your plan includes, then removed automatically. Close your account and everything goes within 30 days.
Questions
How long does it take to set an agent up?
Two to five minutes. You answer a short form, connect what the agent needs, and choose whether it may act on its own.
What does an agent actually read?
Only what you connect: your business profile, a forwarded email, your own web pages, a Google Sheet you choose, or your Search Console data. Nothing else.
Can it send something I have not seen?
Only if you switch that agent to automatic. Until then every email, reply or change waits in your approval queue.
What happens when an agent gets something wrong?
You reject the draft and it is never sent. The activity log shows every step it took, so you can tighten the agent's instructions.
How much does the AI itself cost me?
Nothing extra. Your plan covers a number of runs a month, and a run is one completed job.
What happens if I use up my runs?
Agents pause until the next month, or you upgrade. You are never charged more without asking.
See it on your own business
The free plan gives you one agent and 100 runs a month, with no card.