Key points
- With ten to forty-nine staff, several people may already use AI tools alone, and no one is responsible for AI. The question is how to turn scattered trials into one project that earns its keep.
- Government research found thirty-four per cent of firms this size used AI in twenty twenty-six. Among micro, small and medium firms surveyed, all with five or more staff, the share nearly doubled from twenty twenty-five.
- They say AI use usually means general assistants and AI in software firms already pay for, and a next step suiting many is linking those tools to their own files or systems, to do whole jobs, not just drafts.
- They say the hard part is seldom the AI itself, but settling how a process should work, finding where information is kept, and choosing who checks results. Their example: a twenty-person accountancy firm must agree a standard letter before AI drafts its letters.
- They suggest asking staff what AI tools they use; some will probably be on personal accounts, which they call normal and a sign of demand. Then pick one task several staff repeat, and time it before and a month after any change.
- Among AI users this size, government research found mean total AI spending of ten thousand five hundred pounds in twenty twenty-five: not a project price. They note big spenders lift averages, so many will spend much less. Forty-three per cent of AI users of all sizes spent nothing, which the research links to built-in AI and free tools.
- They point to free government-backed AI courses and subsidised management training; check current terms before applying. Their full guide, which they say no one paid for, is on AI Consultancy Finder.
Transcript
With ten to forty-nine staff, several people may already use AI tools alone, and no one is responsible for AI. The question is how to turn scattered trials into one project that earns its keep. Government research found thirty-four per cent of firms this size used AI in twenty twenty-six.
Among micro, small and medium firms surveyed, all with five or more staff, the share nearly doubled from twenty twenty-five. They say AI use usually means general assistants and AI in software firms already pay for, and a next step suiting many is linking those tools to their own files or systems, to do whole jobs, not just drafts.
They say the hard part is seldom the AI itself, but settling how a process should work, finding where information is kept, and choosing who checks results. Their example: a twenty-person accountancy firm must agree a standard letter before AI drafts its letters.
They suggest asking staff what AI tools they use; some will probably be on personal accounts, which they call normal and a sign of demand. Then pick one task several staff repeat, and time it before and a month after any change. Among AI users this size, government research found mean total AI spending of ten thousand five hundred pounds in twenty twenty-five: not a project price.
They note big spenders lift averages, so many will spend much less. Forty-three per cent of AI users of all sizes spent nothing, which the research links to built-in AI and free tools. They point to free government-backed AI courses and subsidised management training; check current terms before applying.
Their full guide, which they say no one paid for, is on AI Consultancy Finder.