AI is already changing council work. It is drafting correspondence, summarising submissions, finding information and helping staff work through large documents. The question is no longer whether councils will use it. The question is whether people know when to use it, how to check it, and when human judgement must take over.
What is actually changing in council work?
AI is reaching consenting, contact centres, planning, finance, libraries and every corporate function, well beyond anything IT can run as one project. A consent officer may use it to organise application material. A contact centre adviser may use it to find the right policy faster. A finance team may use it to explain a variance or test a draft report.
The work does not disappear. Parts of it shift. Finding, sorting and drafting become faster. Checking, interpreting and explaining become more important. Staff need to recognise weak source material, protect personal information, spot a confident but wrong answer, and explain how a decision was reached. Those are skills, and no software setting supplies them.
Why do "tool-first" pilots stall?
Many pilots begin with a licence, a small user group and a demonstration of what the tool can do. They rarely begin with a clear view of the skills already present in each role. Without that baseline, training becomes generic. Confident users move quickly, cautious users wait, and managers cannot tell whether the pilot improved the work or simply produced more drafts.
A role-level view matters because the risks are different. Using AI to draft an internal meeting summary is not the same as using it to support regulatory advice or communicate with a resident. Staff also need room to say they are unsure. If fear about job change, privacy or getting it wrong stays unspoken, people will either avoid the tool or use it quietly without enough review.
What does a skills first approach look like?
Start with the work, then choose the technology. The useful question for an executive team is which workflows and skills, in which roles, will matter this year. Which AI tool to buy comes after that. Four steps make that question practical.
AI readiness is not a count of licences. It is knowing which people can use the technology well, where they need support, and which decisions still need a human.
What should your council do this quarter?
Choose roles or workflows where AI is already appearing or likely to change the work soon. Bring together the role holders, their manager, people and capability, technology, privacy and risk. Map the relevant workflows. Mark where AI could assist, what information it would touch, what judgement remains with the employee, and what good review looks like.
Use that map for a short learning cycle based on real work. Give people an approved tool, a clear experiment and time to practise, then review the outputs together. Record which skills helped and where people got stuck. That gives you evidence for policy, learning and investment.
