Menu Book a free audit
Practical20 November 20265 min readFayez Nicolas Chaar

AI automation or a new assistant: how to decide

To choose between AI automation and a new assistant, look at the work, not the technology. If the work follows repeatable steps with clear rules, automation can usually handle it. If it needs judgment, relationships or changing priorities, a person is likely to do it better, and the best answer is often a mix: software for the repetitive parts and a person for the rest.

In short

Automate rules, hire for judgment. Break the workload into tasks, sort them into "follows a rule" and "needs a person", and see which pile is larger. That tells you more than any comparison of costs or tools.

How do you split the work?

Start by listing what you would actually hand to a new assistant. Not a job title, but the tasks: answering routine emails, entering data, scheduling, chasing documents, preparing quotes, updating records, looking after clients. Then sort each task:

  • Rule-based: the same steps each time, with clear inputs and a clear definition of done.
  • Pattern-based: mostly repeatable, but with some variation, such as sorting messages or drafting a first reply. AI can help here, usually with a person checking the output.
  • Judgment-based: depends on context, tone, negotiation, priorities or knowing the customer.

If you want help scoring the tasks, what to automate first gives a simple method using time and risk.

When does automation fit better?

  • The work repeats often and is the same each time, such as copying data between systems or sending confirmations.
  • It needs to happen at any hour or immediately, such as acknowledging an enquiry.
  • Consistency matters. Software applies the same rule every time, where people have off days.
  • The work is dull and error-prone for a person, such as retyping details.
  • The volume goes up and down. Software copes with busy periods without recruitment.

When does a person fit better?

  • The work depends on judgment and context that is hard to write as rules.
  • Relationships matter. Customers who are upset or making big decisions usually want a human.
  • The work is varied. If every case is different, there is little to automate.
  • The process is not settled. If you are still working out how it should go, a person can adapt faster than a workflow.
  • The task involves physical presence or things software cannot reach.

Where the result affects a customer directly, be careful about leaving it entirely to software. Our article on AI chatbots for business covers where automated conversation helps and where it fails.

Why is it often a mix?

The choice is rarely all or nothing. A common pattern is to automate the repetitive preparation and give a person the decisions. For example, software gathers the details of a new enquiry, checks them, drafts a reply and puts it in a queue, and the assistant reviews it and handles the conversation. The person spends their time on the work that needs them. See designing a review queue for how that handoff can work.

An assistant can also be the owner of the automation: the person who watches the alerts, handles the exceptions and notices when something looks wrong. Every automation needs someone like that; see why automations break. So automation does not always replace a role. Sometimes it changes what the role does.

What else should you compare?

Think about these points honestly, using your own figures rather than assumptions:

  • Setup effort. Automation takes time to define, build and test before it helps. A new assistant also takes time to recruit and train.
  • Ongoing effort. Automations need maintenance and an owner. People need management, holidays and cover.
  • Flexibility. A person can handle a new situation tomorrow. A workflow needs to be changed.
  • Risk. What happens if it goes wrong, and how quickly would you notice? See the exception problem.
  • Reversibility. A small automation is easy to switch off. A hire is a bigger commitment.

We deliberately do not put numbers on this, because they depend on your business. Our guide to what common automations cost explains the cost factors, and the right comparison is always against your own current cost of doing the work.

A simple way to decide

  1. List the tasks the extra help would cover.
  2. Sort each as rule-based, pattern-based or judgment-based.
  3. If most are rule-based, look at automation first, starting with one process.
  4. If most need judgment, hire, and consider automating the repetitive edges later.
  5. If it is mixed, which is common, automate the repetitive part first and see how much of the workload remains.
  6. Check your processes are written down before either route; see document the process before you automate it.

If you decide to bring in outside help to build the automation, hiring an automation specialist lists what to ask.

Frequently asked questions

Can AI replace an assistant?

It can take over parts of the work, especially repetitive and pattern-based tasks. It is less suited to judgment, relationships and situations that are new, and it usually needs a person to check or handle exceptions.

Should I automate first and hire later?

Often it makes sense to automate the repetitive parts first, because that shows how much work remains and what kind of person you need. But if the workload is mostly judgment, do not delay hiring.

What if I try automation and it does not work?

A small, well-chosen first project limits what you lose, and you can switch it off. That is one reason to start small.

Do I need an agency for this decision?

No. The sorting exercise above is something you can do yourself. Outside help is more useful for the build.

Not sure which way to go? Book a free audit and we will look at the work with you and say honestly whether automation fits.

Want this built for your business?

We will look at how your work flows today and tell you which steps are worth automating, and which are not. Free, and you keep the notes.

You will talk to Fayez or Kareem, the two founders.