Tok's IT Consultant AI automation for small businesses

AI automation for UK small businesses The work nobody hired you to do.

Practical AI automation for UK small businesses, built and handed over by an engineer who runs businesses too.

Get a free automation review

Every week, the admin gets done first.

Sorting the inbox. Copying figures from one system into another. Chasing an invoice, a signature, a reply. Re-keying an order. Building the same report every Monday morning.

None of it is why you started the business, and none of it is what customers pay for. It still has to be done, accurately, every week, and it usually lands on the people you can least spare.

That is exactly the kind of work automation is good at: repetitive, rule-based, high-volume and easy to check. The point is not to replace your people. It is to hand their week back for the work only they can do.

38%

of UK businesses with 10 or more staff that use AI say administrative and clerical roles are the ones most affected.1

A small office desk mid-morning: a laptop, a loose pile of invoices and envelopes, a phone and a mug of tea.
Mid-morning, and the real work has not started yet.

Now add it up.

Admin bill

Prepared for your business
Today

£
Working weeks a yearAssumed, allowing for holidays
46
Hours a year
828
Total a year
£20,700

Example figures. Change any line to make it yours.

That is what admin already costs, every year, before anything changes.

The same desk, late. Every hour on the bill is already paid for.
The same desk the next morning, cleared: one closed folder, a closed laptop and a fresh mug of tea.
The next morning.

What automation hands back.

Automation does not make the work disappear. It moves the repetitive part to software that does it the same way every time, and leaves the judgement with you.

  1. Hours, every week

    Sorting, filing, copying and chasing run on their own, so the day starts on real work.

  2. Fewer mistakes

    A workflow handles the same figure the same way every time, and flags what does not match instead of guessing.

  3. Faster replies

    New enquiries are acknowledged, sorted and routed in minutes, evenings and weekends included.

  4. Growth without another admin hire

    When volume doubles, the workflow does twice the work. Admin stops setting the pace of hiring.

  5. A clear record

    Every run is logged, so you can see what happened, when, and why.

In a study of 5,179 customer support agents, those given an AI assistant resolved 14% more issues per hour on average, and newer staff 34% more.2

Typical first automations

Small, well-defined jobs that pay back quickly. The ones you ticked in the bill come first.

  1. Inbox triageYou ticked this

    New mail is read, labelled and filed. Urgent messages are flagged, routine ones get a drafted reply for you to approve.

  2. Invoice checksYou ticked this

    Supplier invoices are read and matched against orders or agreements. Anything that does not add up is flagged before you pay.

  3. Quote draftsYou ticked this

    Details from an enquiry become a first-draft quote in your own template, ready for a human check.

  4. Bookings and remindersYou ticked this

    Appointments booked from a link, with confirmations and reminders sent automatically.

  5. No more re-keyingYou ticked this

    Information captured once flows into your CRM, accounts and spreadsheets.

  6. Polite chasersYou ticked this

    Unpaid invoices and unanswered quotes get timed, friendly follow-ups.

  7. Monday reportsYou ticked this

    Figures pulled from your systems into one summary, in your inbox before you are.

What changed.

This is now within reach of a small business. Not long ago, automating admin that needs any judgement meant hiring developers. The tools, the costs and the capabilities have all moved since.

  1. 30%

    of UK businesses now use at least one AI technology, up from 9% when the ONS first asked in September 2023.1

  2. 29%

    of businesses with fewer than 10 employees use AI, against 51% of those with 250 or more.1 Large firms have teams for this. Small ones usually do not, which is the gap this service exists to close.

  3. 280×

    cheaper. The cost of running a model at the level of the original ChatGPT fell more than 280-fold between November 2022 and October 2024.3 The cost of the AI inside an automation is no longer the obstacle it was.

  4. 6.4mo

    is about how often the length of task AI agents can complete on their own has doubled since 2019, and faster since 2023, measured on software tasks.4 AI can now carry out a sequence of steps, not just answer a question.

  5. 60%

    of UK firms surveyed named limited AI skills, expertise and knowledge as a barrier, and 71% a lack of identified need.5 Finding the need and supplying the skills is exactly the job.

What that means in practice

  • AI reads messy documents. Invoices, emails, PDFs and photos of receipts become structured data, with a person checking where it matters.
  • Agents take steps. Read the email, look up the customer, draft the reply, file it, log it.
  • Your tools already connect. Gmail, Outlook, Xero, Google Sheets, Shopify and most CRMs can be joined up by workflow tools such as n8n, without replacing anything.
An independent shop counter at opening time with a card reader, parcels wrapped in brown paper and a pot of pens.
Opening time. The admin starts before the first customer.

Sources

  1. Office for National Statistics, Business insights and impact on the UK economy: 8 October 2026. Business Insights and Conditions Survey, September 2026. Official statistics in development.
  2. Brynjolfsson, Li and Raymond, Generative AI at Work, NBER Working Paper 31161; published in the Quarterly Journal of Economics, 2025.
  3. Stanford HAI, AI Index Report 2025: inference cost for GPT-3.5-level performance, November 2022 to October 2024.
  4. METR, Time horizon 1.1, 29 January 2026: doubling about every 196 days from 2019 and about every 131 days since 2023.
  5. Department for Science, Innovation and Technology, AI adoption research, 28 January 2026. Survey of 3,500 UK private firms with 5 or more staff, February to May 2025.

Founder and engineer, Tok's IT Consultant

Built by an engineer who works with AI.

I'm Tok. You speak to me, and I build what we agree. There is no account manager in between.

I'm a software and data engineer with a degree in computer and communication engineering. I have built backend systems, web and iOS apps, and the data pipelines that large organisations run on, including contract work for global data and pharmaceutical firms through this company.

Now I build with AI: agents that read and file email, pull dates into a calendar and research the web, connected to language and image models through their APIs. I know where AI is reliable, where it needs a human check, and how to connect it safely to the tools a business already uses.

I also run my own small businesses, and the first automations I built were for them. Two of them are on the work page.

Engineer and owner in one
Business sense, and the technical depth to build it properly.
Numbers first
Every proposal states what it costs and the hours it should save, before you commit.
You own it
Built on your accounts and documented in plain English. No lock-in.
Careful with data
Each automation gets only the access it needs, designed with UK GDPR in mind.

How a project runs.

  1. Free automation review

    You fill in a short form and we have a 30-minute call about where the time goes. I come back with the two or three automations I would start with and what each should save. No cost, no obligation.

  2. Fixed-scope proposal

    What will be built, what it costs, how long it takes and the hours it should save, in writing. Nothing starts until you approve it.

  3. Build and test

    Built on your own accounts and tested on real examples. You check the results before anything goes live.

  4. Handover and support

    Plain-English notes, a walkthrough for your team, and support while it beds in.

Questions people ask

Do I need to change my software?

Usually not. Automations connect the tools you already use, such as Gmail or Outlook, Xero, Google Sheets, Shopify or your CRM.

Who owns what you build?

You do. It runs on your accounts, and you get the documentation.

What happens to my data?

It stays in your own systems wherever possible, each automation gets only the access it needs, and I use AI services whose business terms do not train models on your data.

How much does it cost?

Every project is fixed price and agreed up front. The free review tells you what I would build and roughly what it would cost before you commit to anything.

What if something breaks?

Every run is logged and failures send me an alert, so problems are caught early. Support is included while it beds in.

Your admin bill: £20,700 a year Get a free automation review