Docket Academy

AI-literacy training for local government

Your staff already use AI.
Do they know when it’s wrong?

Docket Academy is a 30–45 minute simulation. Your employees work a shift at a fictional town counter with an AI assistant that is sometimes confidently wrong — and you see exactly who caught it.

  • 30–45 minutes
  • Runs in the browser
  • Nothing to install
  • No live AI
Ray Ott Records request — April zoning minutes
QuickAI

Great question! Under public records law you have up to 30 days to respond to any records request, and you can take an additional 30-day extension if needed.

Reference panel · Records Policy

The Town acknowledges a request within 5 business days. There is no 30-day response window in Millbrook’s policy.

Would your clerk catch it? From The Monday Counter.
77% of communities name lack of understanding as the top barrier to AI
10% cite political resistance

The gap is knowledge, not politics.

AI is already inside the software you license, and staff are already pasting resident data into free chatbots — usually with no policy to guide them. Training is the cheapest control you have.

Source: ICMA 2024 survey of 635 local governments.

How a shift works

Real counter work. A helpful assistant. Planted mistakes.

  1. 1

    Work the inbox

    A records request, a water-bill dispute, a permit question, a council memo — the ordinary Monday at a small-town counter.

  2. 2

    Ask the assistant

    The AI drafts, summarizes, and answers. Most of it is useful. Some of it is confidently, plausibly wrong.

  3. 3

    Check the record

    The town’s own policies and ordinances sit in a reference panel. Every mistake can be caught from what’s on screen.

  4. 4

    Make the call

    Scored decisions — send, fix, redact, or refuse — with an explanation of the principle after each one.

What your staff learn

Five skills that keep the town out of the headlines.

Recognize AI that’s already here

Spot the AI built into tools the town already uses — and notice that nobody chose it.

Protect resident data

Never paste a Social Security number or medical detail into a consumer chatbot. Privacy depends on the account, not the brand.

Verify, don’t trust

Catch the invented deadline, fee, or ordinance citation before it goes out under the town’s name.

Prompt well

Give a role and a goal, be specific, and check what comes back.

Know when not to use AI

Some answers have to come from a person or an official record — like who gets their water shut off.

What you get as the manager

See where the risk is — before it becomes a headline.

  • Team dashboard. Every person’s score, decisions, and what they asked the assistant.
  • Competency heatmap. At a glance, see that your team got the data-privacy decisions right only 25% of the time — so you know where a workshop would help.
  • Before-and-after scores to show a measurable lift, plus results you can export.

Clerk’s office · 8 staff · The Monday Counter Sample data

Recognize AI that’s already here21/24 correct
88% correct
Protect resident data6/24 correct
25% correct
Verify, don’t trust17/24 correct
71% correct
Prompt well19/24 correct
79% correct
Know when not to use AI15/24 correct
63% correct
Aggregated across every scored decision — no individual is named on the heatmap.

The shifts

Start with the counter. Grow into the hard calls.

Beginner

The Monday Counter

A new clerk clears a Monday inbox: a records request, a water-bill dispute, a permit question, and a council memo.

The AI gets wrong: a records deadline, where resident data goes, an attendance figure.

Intermediate

Permit Season

A permit tech answers zoning questions and reviews an application with an AI pre-screen.

The AI gets wrong: a setback rule, a skipped flood-zone check, an invented ordinance citation.

Intermediate

The Shut-Off List

A billing clerk works from an AI-ranked list of delinquent accounts headed for disconnection.

The AI gets wrong: a biased ranking, a life-support account on the list, a reconnection fee.

Why it holds up

Built so the score means something.

Drawn from real failures

Every planted mistake is modeled on documented cases — invented legal citations, government chatbots that gave illegal advice, biased risk scores.

Every answer is provable

Each scored decision can be settled from what the player was shown. No trick questions, no outside knowledge required.

No live AI, no surprises

The assistant’s replies are written in advance, so every employee meets the same lesson — and resident data never goes to a chatbot.

Bring it to your team

Run a pilot with one department.

Tell us a little about your town. We’ll set up a team, send your staff a join code, and walk you through the results when they’re done.

Want to see it first? Play a shift yourself — it’s free.

Questions managers ask

Who is it for?

Frontline municipal staff — clerks, permit techs, billing and front-desk employees. No technical background is needed.

Does it use real AI? Does any of our data go anywhere?

No. The assistant’s replies are authored in advance and the scenarios use a fictional town. Players never enter real resident data.

What do I need to install?

Nothing. It runs in any modern browser. Staff sign in with an emailed link — no passwords.

How long does it take?

One shift takes 30–45 minutes and can be paused and resumed.

What do I see as a manager?

Each person’s score and decisions, a competency heatmap for the team, and exportable results.

Find out where your team actually stands.