In reality, a community member in a typical US city often has to complete paper forms or fill in online PDFs, and where services are digital, they are inconsistent and still require far too many complex steps. The digital transformation of local government is a multi-trillion-dollar opportunity still waiting to be fully realized. Might artificial intelligence (AI), and specifically AI agents, finally provide the leg up cities need?

Cities Embrace Artificial Intelligence (AI)
It won’t come as a surprise that AI is beginning to find a welcome home in city halls across the world just as it has in every other industry. According to the Hoover Institution, already 1 in 4 government employees regularly use generative AI for their work. That usage level will grow quickly over the next few months following similar trends in the private sector.

AI is finding its way into every aspect of city operations including public safety, planning, transportation, and citizen services. The most popular uses include task automation, support for decision-making, and engagement with the community.

City leaders are recognizing the broader opportunity with AI and are largely embracing it. That said, they currently face significant challenges from their own bureaucracies, regulations, and lack of technical expertise, to risks such as privacy and hallucinations that don’t have a resolution yet. Most limitations, however, are temporary and soon city leaders and providers will find greater ease and more demand for implementing AI-powered solutions.

AI Agents Arrive On The Scene
Perhaps the emerging AI technology that promises the most radical shift in how people experience their local government will be through the deployment of AI agents. An AI agent is a system that acts independently to process information and then take steps to achieve specific objectives. Rather than a person providing AI with the exact actions required to get something done, the promise of an AI agent is that it can determine the optimal steps and then go about getting them done.

OpenAI’s new solution, Operator, is an example of a generalized AI agent. Ask it to find your preferred seats for an upcoming concert and make the booking on your behalf and off it goes.

This, of course, is just a simple tease at what will be possible in the near future when, for example, AI agents coupled with robots will autonomously carry out the entirety of complex assignments.
Transforming The Government Experience
It’s still early for AI agents in the private sector and even earlier for them in public agencies. However, one solution, SuperCity AI, provides an early glimpse at what is coming soon to our cities.
SuperCity is an app that is rethinking how AI can be used to provide a better experience in how residents engage with their city in areas such as finding information, paying bills, and reporting an issue.
Apps that play in this space are already numerous, from SeeClickFix to Nextdoor, and many attempts have been made to hit the sweet spot of convenience and stickiness.
Cities often provide their own solution in addition to competing with offerings from the private sector. The proliferation of community engagement apps for a single city alone creates confusion when people don’t know what to use for a given service, but more broadly, these apps with few exceptions have failed to meet expectations.
The team behind SuperCity come with significant government and technology credentials. Miguel Gamiño Jr., no stranger to city management having served previously as the head of technology in the cities of El Paso, San Francisco, and New York, has joined forces with his two partners, David Lara, formerly the Chief Administrative Officer at New York City Hall, and Niko Dubovsky, who’s worked in the startup world for several years.
The team’s passion for public service together with a deep understanding of how cities work are assets that they are bringing to building this solution. This coupled with state-of-the-art AI adoption doesn’t guarantee their success but certainly provides them with some early advantages.The SuperCity founding team. From Left to Right: Niko Dubovsky, Miguel Gamiño Jr., David Lara.SuperCity
Their mission with SuperCity is to provide a secure and private digital one-stop-shop for residents and to use AI to reduce different elements of friction between the user, the app, and city hall. That friction ranges from residents who are overwhelmed with unnecessary notifications to the complexity of supporting the required interfaces with agency systems. For example, rather than the city being required to manage the complex integration of accepting payments from the app for say, a parking ticket, SuperCity uses AI to meet city requirements and then seamlessly log in and submit the payment.
Removing the complexity for both the user and the city also means that this single app can be used in different cities without requiring the user to download a new app with an entirely different process.
While most apps require the user to locate the feature they need, SuperCity will soon present itself as a conversational bot. A resident will simply discuss what they need and the app will use AI agents to carry out as much of the need with little, if any, user engagement.
Conversational bots are already one of the most popular uses of AI across industries in the area of customer service. Could they also be the future interface for most city interactions too?
The Urgent Future Of AI In Cities
As impressive as the last two years have been, cities are trailing the private sector by a large margin in moving from experimentation to adoption of AI across their functions.
From time to time, a new technology arrives that has the power to radically upset the status quo in a positive way. AI for cities provides perhaps a once in a lifetime shift that will alter what cities do and how they function. City leaders need to increase the urgency of their AI efforts and ensure they are allocating appropriate resources and skills.
In the short-term there are opportunities to have AI augment and improve current operations from community-facing services to data-driven decision-making. Longer term, AI agents will complete entire city services with little or no human interaction on the backend. It’s possible too, that sooner than later, AI will usher in an era without the need for websites and apps.
As the SuperCity app demonstrates, AI and AI agents coupled with novel ideas offer city leaders a whole new toolkit full of possibilities. The time to define an AI future for cities is now.

I turned to Thomas Howell, the founder and CEO of Forum Education, New York City’s leading private tutoring company, to explore what these challenges might be. Forum Education provides best-in-class support to students from elementary school through Law and Medical School. Only 2% of applicants are accepted as tutors, and top instructors can earn $1 million per year.

The Best Teachers Quickly Identify Student Challenges
The hallmark of a good instructor is the ability to identify a student’s source of confusion quickly. Anyone who has taught a mathematics course has encountered a student struggling with a concept and being confused to the point where they cannot articulate a question. In such situations, the best diagnostic tool is not dialogue but observation. By observing how a student starts and stops while working through an exercise, an expert instructor will often immediately pinpoint the issue and begin to address it.

This is just one example of a wide range of nonverbal behavioral signals that the best tutors intuitively recognize. Others include noticing when a student is ready to engage in productive learning versus when they are merely going through the motions to please their parents or when a student finds a particular aspect of a problem interesting while the broader context in which it is being learned is not. The best instructors are attuned to these distinctions and seamlessly adapt to accommodate them.

This is also why traditional assessments become superfluous in one-on-one instruction. A skilled instructor observes a student’s real-time thinking process in a one-on-one context, eliminating much of the guesswork that standardized tests or written assessments are designed to uncover. While the AI can parse text or spoken queries, it remains a struggle to interpret subtle cues like hesitation, tone, or body language—signals an expert human tutor relies on to pinpoint confusion.

The Best Teachers Excel At Motivating Students
Beyond identifying confusion, human tutors excel at another essential aspect of learning: motivation. Motivation is a key component of any instructional project, and an expert instructor needs to inspire her students so that, given the choice between doing the work and doing something with higher short-term appeal, the student chooses to do the work. This often involves fostering in the student the desire to succeed in order to impress the teacher. Any good coach or teacher naturally takes pride in the success of their students, and good students want to show their teachers how well they have done. But there is a big difference between getting praise from your teacher and getting praise from an AI.

How likely is this to happen with AIs? While AI-based gamification—think Tamagotchis or fitness trackers—can prompt certain behaviors, this type of extrinsic motivation differs from the genuine desire to impress a mentor or coach who cares about your success and with whom you have an emotional connection. While a leaderboard may motivate, the motivation is not driven by a desire to impress the app or the leaderboard itself, but by the people, one knows who are paying attention to the leaderboard.

The Best Teachers Are Fundamentally Relational
The reason the machine instructors fall short compared to the best humans is that education is fundamentally relational. When developing online schools in the 2000s, I often emphasized the importance of preserving natural human interaction over flashy technology, with the exhortation that it was the “schoolness” that was essential, not the “onlineness.” While the technology was necessary since the students were scattered around the world, it wasn’t enough to create an online school where students would want to be. One could easily imagine a student saying, “My online high school is amazing. The teachers are great; the other students are super interesting, and I love spending time with them. The system might crash every once in a while, but overall, this school is amazing.” It is much harder to imagine someone saying, “The teachers are ok, maybe not as good as my old school, and the other students are sort of dull, but the immersive video conferencing technology is so amazing; this is the best school ever.”
The Best Teachers Cultivate Intellectual Effort
Against this backdrop, Howell suggests the most significant risk AI poses is not displacing teachers but eroding genuine intellectual effort. “The more AI replaces human work,” he warns, “the less we learn. AI breaks new ground in automation, which may perpetuate the notion that humans can be eliminated from the learning loop altogether.”
The question we must confront is, why are we studying anything? Is it to gain specific knowledge, or is it to develop general skills and build intellectual muscle? If the latter, then if AI does the heavy lifting, students may lose the very practice that cultivates analytical skills. It may be that the presence of AI necessitates an evolution of the curriculum. Still, ultimately, there will remain a proper sequence of exercises to be done to develop the intellect. This is the student’s job; the instructor’s role is to motivate, chastise, celebrate, and optimize.
The Best Teachers Have A Theory Of Mind
Ultimately, what is essential for an expert instructor is a theory of mind. Unless an instructor understands that students possess their own thoughts, beliefs, desires, and emotions—and, more importantly, can discern what those are for each individual student—the instructor will not be able to perform at the highest level. The expertise required is not just in the subject matter but also in understanding the students themselves. This insight cannot be extracted from language, nor does it stem from a semantic model of the subject. Until AI can truly adopt a theory of mind—understanding and responding to students’ individual beliefs, motivations, and emotions—it will remain a powerful tool but not a complete solution. Top-shelf education demands the human touch. The optimal path will remain to combine AI’s efficiency with the empathetic, relational power of the best teachers.