AI pipe dreams come true?
Cities aren’t shy about AI … Street racers (and pipe leaks) beware … And check out the AI Russian nesting doll.
Hello from the A.I. Agenda desk.
I’m Adi, your host.
As the direct manager of a half-dozen AI Agents1, I spend a lot of time talking to computers pretending to be humans.
So I was excited to spend some time this week talking with Tucson City Councilwoman Nikki Lee, a real person who’s at the forefront of implementing AI technologies in government.
As a former Air Force veteran with a background in technology and cybersecurity, Lee is one of the few technically savvy and technologically forward-looking city council people in Arizona, and she’s the co-chair of the National League of Cities’ AI Advisory Committee.
I was excited to ask her about some of the ideas she’s been pushing not only in Tucson, but for other cities of the nation as well.
We’ve uploaded the full interview for our paid subscribers. If you want to get in on all sorts of AI conversations with officials, upgrade today!
From preventing costly water main breaks to curbing street racing and even co-writing the city’s new General Plan, AI is doing some rather down-to-earth work in the desert.
AI in your pipes
Tucson taxpayers shell out roughly $1 billion every decade to repair the city’s 4,600 miles of water pipelines.
Lee says one of the things the city is trying to do with AI is get out of a reactive mode when it comes to repairing those pipes. Instead of waiting until a pipe bursts and water pours onto Tucson’s streets, which is expensive and a huge inconvenience to daily commuters, city officials want to use AI tools to predict which pipes are going to break and then replace them before the pipes actually burst.
To get a better handle on these pricey mishaps, the city teamed up with Voda.ai, whose AI scours historical break data, pipe age, and work orders to flag the “likelihood of failure” along each segment.
“It definitely helps us be really smart in applying capital investment dollars,” Lee told us.
One thing city officials can’t predict, however, is contractors accidentally breaking pipes during construction projects, which Lee says happens more often than you might expect.
The city’s AI effort with water pipes is pretty new. City officials haven’t disclosed the tech’s price tag or how much savings they’re expecting to see.
AI takes on street racers
Tucson officials also are using AI to tackle one of the highest-profile public safety issues in the city: street racers.
Street races have been a big problem in Tucson since the early days of the pandemic. City officials have tried for more than a year to put a lid on racers who take over streets, do donuts, light fireworks, and generally create mayhem.
“There's so many shenanigans happening on the roads out there. It's crazy down here,” Lee said.
The city is using AI as part of a larger effort to crack down on street racers. Lee said she couldn’t share too many of the specifics, saying it was a law enforcement operation and she didn’t want to give away the details.
What she could tell us is the city launched a response center they can set in motion whenever they get a call about street racers. The center pulls together large amounts of data, including from license plate readers and bodycams. Some of those bodycameras are set up without the “body,” placed in locations that police want to monitor (even polling locations during elections).2
Yes, AI helped draft Plan Tucson 2025
For the past few months, Tucson officials have been reaching out to the public to get feedback on the city’s master plan, which will go to voters later this year.
That led to a lot of feedback on Plan Tucson 2025 and city officials had to find a way to sort through it all.
City officials used generative AI tools like ChatGPT and Claude to analyze transcripts from community meetings, surveys, and open forums to find the recurring themes – like sustainable growth, equitable housing, and climate resilience – that became the backbone of the draft.
The AI wasn’t drafting the plan wholesale.3 Instead, it acted as a data-mining assistant, clustering feedback into actionable insights that planners could incorporate.
For example, community comments about walkable neighborhoods or water conservation were grouped into broader categories, helping planners see patterns at scale.
Tucson’s team stresses that while AI streamlined the process, the decisions about what made it into the plan were firmly in human hands.
This isn’t a closed book: residents can still review and comment on the draft at PlanTucson.org. If you’ve ever wanted to critique both city planning and AI’s role in it, now’s your chance.
It’s not just Tucson
Tucson isn’t alone in exploring AI. Across the state, different municipalities are weaving AI into local governance – each with its own flavor:
Phoenix uses AI to optimize traffic signals in real-time and monitor wastewater systems. It also partners with The Connective, a regional consortium for “smart city” initiatives that have projects ranging from VR/XR tourism to civics engagement apps.
Tempe adopted an Ethical AI Policy to ensure fairness and transparency before rolling out large-scale AI projects. By putting the framework first, the city hopes to dodge biases within AI and avoid privacy mishaps when it eventually deploys chatbots or automated tools.
Gilbert appointed a Chief AI Strategy and Transformation Officer and set up an AI committee. It also partners with private tech firms to extend into cybersecurity and data governance.
Cities elsewhere in the country are doing even more interesting stuff:4
Dearborn, Michigan, is using AI to offer translation services in a city where half of residents don’t speak English at home.
Washington, D.C. is using AI chatbots to simplify interactions with the city’s open data.
Memphis, Tennessee, is using AI to detect potholes by analyzing video from city vehicles. It also flags urban blight.
So…where does this lead?
These initiatives show that Arizona cities aim to solve concrete problems – water waste, traffic woes, public safety – not build futuristic robot armies.
Some of these pilot programs will flourish, others may fizzle. But with collaborative coalitions, early ethics checks, and a willingness to experiment, these futuristic AI projects might well become tomorrow’s routine public services.
Tracking all the bills introduced at the Arizona Legislature can be exhausting. That’s why we built Skywolf, our AI-powered legislative tracking tool. It’s designed to sift through the chaos of state legislation so you can focus on what matters.
Bill Tracking: Track each bill through each stage of its life cycle and receive text, email or push notifications any time something happens to that bill.
Deep Insights: Receive AI-generated, human-vetted expert summaries of each bill introduced and amendments.
Customizable Reports: Group bills into tracking lists to cover all the action on any given topic and receive customizable reports on their status.
Grassroots Advocacy Tools: A unique email generator to empower grassroots lobbying.
We’re watching, so you don’t have to scroll endlessly through legislative PDFs.
Sometimes AI tools are intimidating, but Lovable.dev lives up to its name. This charming platform is designed to help non-developers create interactive tools and websites that don’t just work – they delight.
As an example of what can be done, here’s a sneak peek of a tool we are starting to put together for our subscribers ;)
We are using Lovable’s AI Agent to build an interactive dashboard to take in data from more than a dozen sources about AI legislation across the U.S. and give a beautiful dashboard view of it.
Here’s a meta moment for you: this week’s newsletter was brought to you in part by Deep Researcher AI, a tool we used to track down AI-related initiatives in cities, dig up legislation on AI ethics, and (ironically) discover AI’s role in defining AI itself.
Yes, we’re aware this might mean an AI helped write about AI ethics using data on AI governance policies that were likely influenced by AI tools.
It’s AI all the way down.
We promise: there were humans involved too. At least… we think we were involved.
Until next time, stay curious (and keep asking who’s really writing this stuff).
Fun Fact: An AI agent is a software program designed to make decisions independently to achieve goals set by humans. These agents typically have access to resources such as tools (e.g., Connection to other apps, web crawlers or other functionalities to interact with systems), memory (e.g., the ability to recall previous interactions or knowledge), and storage (e.g., a database or other systems to retain long-term information) to help them effectively complete their tasks.
Tucson readers might be interested to know some of these AI initiatives in local law enforcement are included in Tucson’s Prop 414, the sales tax that heads to voters in March.
Lawmakers have used large language models (LLMs) to draft summaries, analyze public comments, and even prepare legislation on AI deepfakes.