← LeanIT Services Guide

How Do You Know If You Need AI or Automation?

By Lee Gibson, LeanIT Services, Banbury, UK · Last updated 27 August 2026

Yes, AI is worth it for a small business, but only for the right kind of work, used within clear boundaries. AI and automation are not the same tool. Automation carries out repeatable tasks exactly the same way every time, with no interpretation involved. AI interprets, writes, and reasons, inside limits you set and check. Most confusion, and most failed AI projects, come from treating the two as interchangeable.

AI and automation are not the same thing

People use "AI" and "automation" as if they're one thing. They solve different problems, and mixing them up is where a lot of wasted spend comes from.

Automation executes a defined, repeatable task exactly as instructed, with no judgement involved. It is reliable precisely because it does not interpret anything.

AI interprets, drafts, and reasons about language or unstructured information, inside boundaries a person has set, and its output should always be checked, not assumed correct.

A simple example: in a client meeting, automation is what takes down exactly what was said, a verbatim transcript, the same every time, with nothing added or left out. AI is what turns that transcript into a short summary and a list of action points, a judgement call about what mattered. Often the right answer is both, used for what each one is actually good at.

Is a task ready to be automated? The three-question test

This test is specifically for automation candidates. Answer these in order; most businesses can rule themselves in or out within a 30-minute conversation.

  1. Is the task repeatable? If the steps are different every time, automation has nothing consistent to grab onto. Fix the process first.
  2. Do you already own something that solves it? A large share of "we need AI" requests are actually "we're not using the software we already pay for properly." Check this before building anything new.
  3. Can you put a number on the return? Hours saved per week, errors avoided, or revenue unblocked. If nobody can estimate the return, it's not ready to be built; it's still a discovery conversation.

Manual fix, existing software, automation, or AI

SituationWhat actually solves it
The process is inconsistent, undocumented, or different depending on who does itMap and standardise the process first, no technology yet
The task is repeatable but the tools you already pay for aren't set up to handle itReconfigure or better use existing software
The task is repeatable, well-documented, and provably costs time or introduces risk when done by handA custom-built automation, hosted and monitored long-term
The task involves interpreting, drafting, or summarising unstructured information, not just repeating stepsAI, used inside a clearly bounded task with a person checking the output
Someone wants "AI" because a competitor has it, with no specific outcome namedNothing yet, start with the outcome, not the technology

When not to use AI

AI is not appropriate for low-volume, judgement-heavy, or constantly changing work, since the effort of setting and maintaining safe boundaries around it outweighs the time it saves. It is also not appropriate as a first move on a process nobody has actually observed end to end, because letting AI or automation loose on an unmapped process just makes existing mistakes happen faster and with less visibility.

How to implement it safely, with visibility

Once a task is genuinely a fit for AI or automation, the harder question is how to run it without losing sight of what it's doing. Four things matter most:

  1. Set the boundary before you switch it on. Write down exactly what the AI or automation is allowed to touch, decide, or send, and what always stays with a person. A boundary nobody wrote down is a boundary that will eventually get crossed.
  2. Keep a human checking the output, not just the process. Automation rarely needs checking once it's built and tested, because it does the same thing every time. AI output should be checked on a schedule that matches how much it matters, every time at first, sampled once it's proven reliable.
  3. Log what ran, when, and on what data. An automation with no audit trail is a black box the moment something goes wrong. Every automated step should leave a record that a person can read back later without needing to ask the system what it did.
  4. Monitor it after launch, not just at handover. A process that was safe on day one can drift as the data, the tools around it, or the task itself changes. This is the reason LeanIT keeps monitoring and managing what it builds rather than handing it over and walking away.

How this is different from hiring an AI consultant

Most "AI consultants" hand over a recommendation and leave the building, hosting, and maintenance to you or your internal team. LeanIT Services does the opposite. Process mapping and the recommendation are the free or fixed-fee front end, but the actual automation, the working software, is designed, built, hosted, and monitored by LeanIT on an ongoing basis, so the client never inherits the technical risk or the maintenance burden. See what we do and how it's priced.

Frequently asked questions

Is AI worth it for a small business?

Yes, but only within clear boundaries and only for the kind of work AI is actually good at: interpreting, drafting, and summarising unstructured information, checked by a person. It is not the right tool for repeatable, rules-based tasks, that's automation's job, and confusing the two is a common reason AI projects disappoint.

What's the difference between AI and automation?

Automation carries out a defined, repeatable task the same way every time, with no interpretation involved, such as capturing an exact transcript of a meeting. AI interprets and reasons about language or unstructured information within limits a person sets, such as turning that same transcript into a summary and action points. They solve different problems and are often used together, not as substitutes for each other.

How much does business process automation cost in the UK?

There's no fixed rate card. It starts with a free 30-minute discovery call, and the price follows the scope that call uncovers. A narrow, well-understood task (like a month-end process run by hand in a spreadsheet) tends toward a lower fixed quote, since most of the pain is easy to see and remove quickly. A wider or less-understood process (like reworking how invoicing works end to end) needs proper mapping first, so the quote is higher to reflect that extra discovery. See pricing.

What's the difference between an AI consultant and an automation build partner?

A consultant typically advises and hands over a recommendation. A build partner like LeanIT Services designs, builds, hosts, and maintains the working automation itself, so the client is leasing a managed outcome rather than managing new software in-house.

How do you implement AI or automation safely, with visibility?

Four things: write down what it's allowed to touch or decide before switching it on, keep a person checking AI output on a schedule that matches how much it matters, log what every automated step did so nothing runs as a black box, and keep monitoring it after launch rather than handing it over and walking away.

Do I need a developer on staff to automate my business processes?

No. Under a managed service model, the build partner owns the infrastructure, hosting, and maintenance, so the client needs no in-house technical staff to keep the automation running.

How long does it take to design a business process automation?

Initial process mapping is designed to move in days, not months: a short discovery conversation, a fast map of the current process and its bottlenecks, and a design review, before any build work is agreed.

Lee Gibson has spent 17 years designing, building, and supporting IT systems and services across large organisations and startups, including service delivery management and complex service blueprinting. He founded LeanIT Services in Banbury, UK to bring that same process discipline to small and medium businesses.

Have a specific process in mind, or just know something could run smoother?

Start a Conversation