Service
AI Automation for Small Businesses
AI is only worth using when it removes real work. I find the jobs in your business that are repetitive, rule-heavy and language-based — then build systems that handle them reliably.
AI automation in plain English
Most business admin is reading something, deciding what it is, and then doing a predictable thing with it. Reading an enquiry and deciding which department it belongs to. Reading an invoice and typing the numbers into accounting software. Reading a booking request and checking whether you have availability.
AI automation is simply software that can do the reading and deciding part, plugged into the systems that do the rest. That's it. No robots, no reinventing your business — just removing the parts of the day that a person does not need to be involved in.
I build these as proper systems with logging, error handling and a human review step wherever the cost of getting it wrong is higher than the cost of checking. You keep control; the software takes the volume.
Where AI genuinely earns its place
These are the categories of work I see paying back fastest in small and medium businesses.
Incoming email classification
Sort the inbox before anyone opens it: enquiry, invoice, supplier, complaint, spam. Route each to the right person, folder or system automatically.
Draft email responses
Common replies drafted in your tone with the right details pulled in, sitting in the drafts folder for someone to check and send.
Customer enquiry handling
Understand what an enquiry is actually asking, retrieve the answer from your own information, and respond or escalate accordingly.
Document and invoice processing
Pull line items, totals, dates and references out of PDFs and attachments, then push them into accounting or a database without retyping.
CRM updates and task generation
Keep records current automatically after calls, emails and form submissions, and create the follow-up tasks nobody remembers to log.
Knowledge assistants and reporting
Let staff ask questions of your own documents and data, and turn raw operational data into the weekly summary someone currently writes by hand.
What could AI automate in your business?
A useful way to think about it: follow one task from the moment it arrives to the moment it's finished, and mark every step a person has to do manually. Those steps are the shortlist. Here are workflows I build regularly.
- Enquiry arrives
- AI understands the request
- Information retrieved
- Reply drafted
- CRM updated
- Follow-up scheduled
- Invoice email received
- PDF read and data extracted
- Matched to purchase order
- Pushed to accounting
- Filed and archived
- Form submitted
- Intent and urgency classified
- Owner assigned
- Personalised reply drafted
- Reminder set if no response
How I work
I start by mapping the workflow rather than the technology. If a process is unclear or inconsistent between staff, automating it just makes the inconsistency faster — so the first job is agreeing what good looks like.
From there I build the smallest useful version and run it alongside your current process. You can see exactly what it would have done before you trust it to do it. Once it's earning its keep, I widen the scope.
Everything connects to the tools you already use. In most projects no existing software is replaced — the AI layer sits between what you have and closes the gaps.
FAQ