AI readiness in local government is not about picking a tool. It is about knowing where you stand, putting a short policy and a named owner in place, and then fixing a few painful workflows you can measure. In the AI readiness workshop Avero ran on October 15, the point was simple: start with the pain, not the tool, and put a baseline and a date on everything.
Your staff are almost certainly using AI already, often on personal accounts. Most agencies have no baseline for where they stand. That gap is where risk and wasted spend live.
How do we know where we stand today?
Use a maturity diagnostic and place your organization on a stage. The workshop used five: foundation, planning, table stakes, applications, and autonomy. Most public organizations sit in the first two, and that is fine. The value is in knowing it.
AV’s view is that this is not a one-time exercise. Retake the diagnostic quarterly. Things move fast, and a quarterly re-baseline shows leadership whether the work is actually moving the organization forward.
What has to be in place before we build anything?
Three things, in order: literacy, policy, and ownership. Quick wins come after. Agents come last.
Literacy. People cannot use AI safely if they do not understand what it does well and where it fails.
Policy. In the workshop, the point was that an AI policy should fit on about two pages. Cover approved tools, prohibited uses, how data is handled, and who to ask. A long policy nobody reads is not a policy. Neither is silence.
Ownership. Name a torchbearer: a trusted person who owns literacy, policy, and adoption. AV’s view is that AI ownership is a leadership job, not an IT ticket. IT is a key partner, but the torchbearer does not have to sit there by default.
How do we keep our data safe?
Confidential data and personal information are the top blockers, and they should be. The practical moves:
- Use enterprise-grade tools with clear data handling terms.
- Separate personal and business accounts. No agency work on personal AI accounts.
- Classify what data can go into which tool, and put that in the two-page policy.
Your data foundation matters here too. If records in your ERP and core systems are inconsistent, AI will surface that fast. Readiness work is often data work.
Which AI projects should we start with?
Write down the three most tedious workflows in your organization. Then, for each one:
- Set a baseline: hours spent, backlog size, error rate.
- Assign an owner.
- Set a target date.
- Define the KPI that proves it worked.
Then pick the tool. AV’s point in the workshop was blunt: without a baseline, there is no ROI to report. Good measures include hours saved, capacity redeployed to higher value work, and service quality for residents.
How do we budget for AI without surprises?
Watch consumption pricing. Usage-based pricing can break budgets that were built on an annual cycle. AV’s view is that consumption pricing is a budget surprise waiting to happen for most public organizations.
Where you can, prefer fixed-fee licensing. Where consumption pricing is unavoidable, set caps, especially for power users, and review usage monthly. Put these criteria into procurement before you sign, not after the first bill.
How do we stop pilots from stalling?
Give every AI project a kill-or-fund date. By that date it either shows value against its baseline, or it stops.
The workshop called out a common pattern: stalled, quickly built apps with no requirements, no security review, no owner, and no KPI. They burn time quietly. A kill-or-fund review cadence keeps the portfolio honest and frees people for work that is paying off. Tie those reviews to a 6 to 12 month roadmap your council or board can adopt.
How do we bring our workforce along?
Role anxiety is real. In the workshop, the point was that if leaders do not explain why the organization is adopting AI, staff will fill the gap with fear.
Explain the why early and often. Frame the goal as redeploying capacity, not just cutting it. Show people what the backlog looks like today and what they could do with the time back.
What does this look like in practice?
One county we work with started exactly this way: data first, then use cases, then a roadmap. The program covered an enterprise data strategy, ranked use cases with ROI cases, vendor and pricing criteria, a governance cadence, and literacy training. The order mattered more than any single tool choice.
For more on the governance side, see AI governance in government: trust requires transparency and AI risk frameworks for public sector leaders. If your pilots are stuck before they start, read the AI pilot that never starts is still a procurement problem.
FAQ
What is an AI readiness assessment? A structured review of your data, systems, security, tools, policy, and workforce literacy that places you on a maturity stage and produces prioritized next steps with owners and dates.
How long should our AI policy be? About two pages. Approved tools, prohibited uses, data handling, and who to ask.
Who should own AI in a local government? A named torchbearer with leadership backing. Not automatically IT.
Want a deeper starting point? Read Chapter 1 of Start at Zero free.