Most companies don't need an AI agent. Here is when you do.

Agentic automation is AI that reads a situation, decides the next step and takes it in your tools. Most of the repeat work in a growing business doesn’t need that. It needs a standing order: a known rule, run every time.

We build both. This is how we decide which one a job gets.

What is agentic automation?

An AI agent is a system that works toward a goal in steps. It reads what came in, chooses what to do, acts in your tools, and checks the result.

Agentic automation is a workflow where one or more of those steps is run by an agent. The rest of the workflow can still be plain rules.

What is a standing order?

A standing order is an instruction that runs without being asked.

Send the report every Friday. Remind the client when an invoice is seven days overdue. Flag any lead from a company over 50 people.

The model can still do the reading and drafting inside a standing order. The rule decides what happens next.

How do you tell which one a job needs?

Write the rule down. If you can say it in one sentence, and it holds for nearly every case, the job needs a standing order.

If the next step depends on reading something and making a call, that step needs an agent. Which reply fits this enquiry? Which of these pages should we fix first?

Question	Standing order	AI agent
Can you write the rule in one sentence?	Yes	No, it depends on what comes in
Is the next step the same every time?	Yes	It changes with each case
What does it cost to run?	Little, and the same each run	Model usage on every run, plus reviews
What does a person check?	The cases the rule doesn’t cover	Anything that sends, hires or moves money
How does it go wrong?	The rule changes and nobody updates it	It acts on a wrong reading

Most jobs are a mix. An invoice reminder is a standing order. Matching a payment that arrived with the wrong reference may need an agent.

Which jobs need an agent, and which don’t?

Job	What it runs on	Why
Publishing a page once it is written (Mockey)	Standing order	The page is pushed to GitHub and goes live. Same step every time.
Fixing site code from an issue (Mockey)	AI agent	A Claude routine reads each open issue and decides what to change. A developer reviews the code.
Choosing which pages to write and fix for SEO (Mockey)	AI agent	It reads Search Console and picks the work worth doing next.
Answering a new lead	AI agent	The reply depends on the enquiry and your past deals. A person approves it before it sends.
The Friday report	Standing order	The numbers and the format are the same each week.
Chasing an overdue invoice	Standing order	The rule is fixed. A person approves anything that moves money.

The three Mockey rows run today. The publishing and code steps are in how Mockey moved 328 posts off WordPress and handed the work to Claude.

Is agentic AI the same as generative AI?

No. Generative AI produces text, images or code when you ask. Agentic AI uses that ability to take steps toward a goal, such as reading an inbox, drafting the replies and filing them.

An agent uses generative AI. A chatbot that only answers questions is generative AI without the agent.

Why do we start with standing orders?

A standing order costs less to run, and there are fewer ways for it to go wrong. When it fails, the reason is easy to find: the rule changed.

An agent earns its cost where the judgement is the slow part of the job. Reading every new enquiry and drafting a reply that fits it is that kind of work.

Where should you start?

List the jobs your team repeats every week. Next to each one, write the rule it follows.

The jobs where you can’t write the rule are the ones worth an agent. Bring that list to the calendar below, and we will tell you which ones we would build first.