There are two ways to get AI working in a small business. Buy a tool, or build one. Both are right in different situations, and the wrong choice is expensive in different ways.
Buy a tool when...
- The job is generic: transcription, meeting notes, copywriting, image editing.
- You need it working this week.
- The data involved is not sensitive or business-critical.
- One or two people will use it.
Off-the-shelf tools are excellent value for generic work. There is no sense paying anyone to rebuild them.
Build when...
- The process is specific to how you work, and no tool matches it without compromise.
- The work spans several systems — your accounts package, your job sheets, your website, your inbox.
- You are paying for four or five subscriptions that still do not talk to each other.
- The output needs to land in your own database, not somebody else's.
- Getting it wrong has a cost: pricing, compliance, safety, customer data.
The hidden cost of tool sprawl
A typical Oxfordshire business we meet is running a CRM, a form builder, a scheduling tool, a quoting tool and a spreadsheet that ties them together by hand. Individually each is affordable. Together they cost more per year than a purpose-built internal app, and the manual re-keying between them is where the real money goes.
The middle path
Most of our work is a hybrid: keep the good off-the-shelf tools, and build the thin connective layer that makes them behave like one system. That layer is where automation earns its keep — and it is far cheaper than replacing everything.
How we decide
We count three things: hours spent, error rate, and how often the process changes. High hours plus stable rules means automate now. High change means keep a human in it and automate the edges first.
If you want a second opinion before signing another annual subscription, we are happy to give one.