Public Sector
Municipal and state buyers live with real constraints: procurement rules, public records, budgets set a year out. We've done AI work inside those constraints, and we scope it the way you have to buy it.
The person who knows why the system is set up that way has a date circled on the calendar. Capture the judgment before the retirement party — the interviews, the documents, the decisions — and make it searchable for whoever inherits the desk.
Permits, records requests, 311 queues. Demand grows; headcount doesn't. The fix isn't a chatbot on the homepage — it's working the queue: what can be answered instantly, drafted automatically, or routed before a human ever touches it.
Your staff didn't wait for a policy. Governance means knowing what's in use, what data it touches, and what's allowed — written down and defensible before the audit or the records request asks.
Engagements this year
Eight government bodies this year. Cities and a county we can name, and state work we describe by area. The work spans operations, knowledge capture, data, and health.
18 custom AI agents at work across city operations, with 40+ staff using them across departments.
Capturing institutional knowledge before it walks out the door. Interviews, documents, and decisions, structured and searchable.
A studio of civic agents in staff hands, working the daily load: the meeting coming up, the notice going out, the question that used to wait on one person.
A county carries a wider mix of duties than a city. Same approach: put the knowledge where the staff already work, and keep it current.
A town's staff wear more hats than a city's. The studio is sized to the people who actually run the place.
An engagement that starts at the leadership table instead of a platform: what AI is for here, and where the first build should land.
Work in the state's data and analytics arena: many agencies, one picture, and the stewardship questions that come with it.
Public health runs on mandated reporting and tight staffing. The information has to be right the first time, at scale.
Roughly 130 people hold seats across the municipal studios, and the problems repeat from one government to the next. Bring yours. Odds are we have already met it.
Read the Fishers case study →Running today
These run in live studios today, built with the departments that asked for them. More than 5,000 people trained across the practice.
One civic agent per city, running on a managed system prompt written against that city's own ordinances, forms, and past practice. It answers the questions staff used to walk down the hall to ask. We keep the prompt current, so the assistant does not drift as policy changes.
The approach: custom platform app, managed prompts
It drafts your AI use policy in your own language: which tools are allowed, what data may go into them, when disclosure is required, who approves an exception. Staff get a rule they can actually follow, and you get one you can defend.
The approach: policy drafting, governance guardrails
Built for the constraints an Indiana finance officer actually works inside: levy limits, tax caps, and a budget calendar nobody can move. It reasons in that context instead of generic public finance, which is the difference between an answer you can take to council and one you cannot.
The approach: domain-tuned advisor, curated knowledge base
The wider map
Each of these started the same way. A department named the problem, and we built the smallest thing that solved it.
Work every agenda item before the meeting: the questions coming, and the answer you should already have.
Check the notice, the executive session, and the records response before somebody files a complaint about them.
Draft the first statement and the internal notice while the incident is still moving, not after.
Score government IT proposals against your requirements side by side, with the gaps written down.
Read a vendor agreement for data rights, training clauses, and the terms that outlive the contract.
See what a bill actually does to your ordinances, your budget, and the staff who carry it.
Turn adopted policy into the steps, forms, and owners that make it real at the counter.
Minutes drafted in your clerk's format, ready for review while the meeting is still fresh.
Rewrite an ordinance or a notice so a resident can read it once and act.
Publish notices in the languages your residents actually speak, without waiting on a translation vendor.
Test a message against constituent personas first, so the objections show up before the mailer does.
Draft replies and newsletters in your office's voice, so the tenth one reads like the first.
Work a contested issue from every side before it reaches public comment and the local paper.
Answers the staff question mid-task, which is when training either sticks or quietly dies.
Turn a vague push to use AI into a scoped problem with a testable outcome.
Match the tool to the job and to your procurement rules, before somebody buys the wrong one.
None of these are roadmap items. They run today in live studios. Bring the item stuck on your desk, and we will say which one fits.
