Practical AI for smaller organisations: where to begin, and where to wait

There is no shortage of enthusiasm about artificial intelligence. For a smaller organisation, the real question is narrower: which tasks, in our business, would genuinely benefit — and what could go wrong?

Good places to start

• Drafting and summarising: first drafts of proposals, meeting notes, internal documentation.
• Classifying incoming information: routing enquiries or tagging documents for review.
• Searching your own knowledge: helping staff find answers in policies and past work.
• Structured data entry: extracting key fields from forms or invoices, with human checking.

These share a common trait: a person stays in the loop, mistakes are easy to spot, and the time saved is real.

Places to be cautious

Be careful wherever outputs go directly to customers without review, where decisions affect people's finances, health or employment, or where confidential data would be sent to a service whose terms you have not read. In these areas, the cost of an error can outweigh the efficiency gained.

A simple approach

• List the repetitive, text-heavy tasks your team does each week.
• Pick one with low risk and a clear measure of time spent.
• Pilot for a few weeks with a small group and written guidelines.
• Decide, based on evidence, whether to expand, adjust or stop.

Treated this way, AI becomes one more well-governed tool in the operating model rather than a source of anxiety or hype.