For those of us in civic tech, the bigger story is inclusivity and accessibility. AI can help us hear constituents who have never been listened to and build technology for institutions that otherwise have neither the money nor the time to build it themselves. For the first time, we’re told building software is something almost anyone can do. Our thing is seeing what that promise looks like in practice.

Once more, together with CUNY PIT Lab and Civic Roundtable, we brought together 80+ technologists who build, practitioners who know the workflows, advocates who ensure equity, and decision makers who use the products. Unlike last time, we gave teams two narrowly scoped themes to see how well they could color inside the lines (with a bit of their own flair, of course).

The first topic was improving community boards. Our own Tal Roded sits on Manhattan’s CB3, and he’s seen firsthand how little tech infrastructure exists for the most local level of community expression in our democracy. We challenged participants to find creative ways to help.

The second came from something any New Yorker has likely noticed: Mayor Mamdani stepping up to fix mundane public nuisance. As much as we admire the Mayor’s dedication to the cause, we figured there had to be a better way for agencies to capture, triage, and act on New Yorkers’ requests. Our second theme, capturing public voices, thus asked teams to generate innovative solutions that might build trust in government by helping the public feel heard.

Our panel of expert judges (hat tip to Jeremy Ney, Abbey Kos, Raj Korpan, Sarah Lind, and Amaury Dujardin!) also pitched their own ideas to the builders, from improving transit-oriented development to centering people and communities in the city’s plans.

Rather than providing prescriptive solutions, we gave teams a direction and let them iterate rapidly on what might actually help. The result, as usual, was a delightful experiment in giving State Capacity enthusiasts real problems, knowledgeable people to learn from, and enough room to surprise us.

Fixing community board operations

It’s a common trope in government that “what to build” can be almost too easy to define, with the resounding answer “everything”; the challenge is knowing where to start. This was absolutely the case for teams working on Community Board tools. With so many critical needs, where to begin?

We were lucky to be joined by a number of community board members-turned-builders, and each tool was designed and built with community board members in mind, including the ones in attendance. Many of the projects were very much of the “scratching your own itch” kind. Prototypes targeted specific, clear needs: fixing broken email-based workflows, serving as (currently non-existent) systems of record for past decisions, and reconciling commitments from members with actual work done by City Hall.

• Budget Requests helps members track priorities from their past years’ Statements of Needs and view agencies’ responses, so they can advocate for their needs with those agencies

• BoardOS was designed to present a representative with every piece of information they needed in order to be prepared for their meetings, helping extricate them from the process of manually reviewing 12 PDFs before each session

• Robert Rules, informed by a CB6 member’s experience arguing about Robert’s Rules of Order, made it easy and fun to teach new members how CB meetings are run

• Boardcast helps community board members spread awareness of their meetings by taking meeting notes and turning them into social media posts

The solutions weren’t complicated, but they were previously out of reach, and the most local form of City government has been hindered by their lack. If we can’t get it right at the foundational level, how do we expect to get it right elsewhere?

Capturing voice to build trust

Our second set of projects tackled whether New Yorkers believe the city hears them. They shared a premise: when residents can see they’re being listened to, they invest in the outcome.

• Hearts and Minds pulls together what’s happening on a given issue in a given place, such as the number of residents in a district who lost SNAP benefits under H.R. 1, and sets it next to the public reaction on social media.

• “Mamdani, Fix This” takes every TikTok asking the mayor for help and routes it to the relevant agency / a public queue, so a viral complaint turns into something trackable.

These tools are even more powerful in concert. Together, they could give the city a running read on public sentiment and offer residents proof that someone is paying attention.

Judge specials: simple solutions for complicated problems

Much of what cities deal with is just plain hard. Rezoning is a good example: turning a laundromat or gas station into housing runs through land use rules, environmental review, neighborhood pushback, and years of process.

In the Zone skipped the urge to model all of it and built a single map showing where density is low across NYC and what would change for residents and planners if those lots were rezoned. The ability to generate an actual image of what the future could look like (a gas station turned into a residential development, for example) was a huge unlock for our judges, who described countless community engagement meetings trying and failing to help the public visualize a future development or project. A classic demonstration that a picture is worth a thousand words.

What’s next

Across all themes, the strongest projects came from teams who leveraged firsthand knowledge: board members tired of PDF stacks, residents who sat through rezoning hearings, and organizers watching complaints pile up on social media with nowhere to go. Two hours was enough to show that there was something worth building.

That is why we love our hackathons. AI has made prototyping software dramatically easier. The harder part is everything around it: figuring out what problem to solve, designing how the tool should be built, and determining who needs to be involved so that the end product matters in the real world. We design these events to start with the need in mind, not the technology.

What’s next for us is simple. We’re growing the ecosystem. That means incubating projects, holding build-oriented events such as demo nights and hackathons, and hosting community-building events such as salons and meet-ups.

What’s next for you is what matters. You are the ecosystem. We need passionate builders who have a civic tech itch to scratch. We need government practitioners who want to learn how to use emerging technology to augment their work. We want philanthropy and VC in the room to think about funding all work State Capacity. We’re looking for visionaries who might take our efforts beyond NYC.

Perhaps most critically, we are always looking to partner with agencies and departments that want to tackle big, nebulous problems where tech just might help.

If any of this sounds like you, reach out.

And if you just want to dip your toe in the water and be the first to hear what we’ve got going on, the easiest place to start is by joining the SCE Slack.

In 2017, Rhode Island’s child welfare agency had an ambitious plan: hire Palantir to use big data to match foster kids to foster homes. They were sure the problem was access to data. Then Marina looked at the database. It was a physical box of cards. Probably not the same data access fix Rhode Island had in mind.

The rest of her essay follows a decade of ongoing fixes. Each fix worked. Each one turned out to be a symptom of another problem. Solving a broken intake process led to state rules that couldn’t be explained. Understanding those rules led to a federal model standard that had lost its guidance on the way down. It took rewriting the national standard itself to fix the problem, and the work is ongoing.

Marina’s description of her work may be long, but it is worth every minute of reading. Once we finished it, the State Capacity Ecosystem team had a practitioner’s question.

How do you actually do all this?

If you’re a well-meaning engineer, product manager, or consultant told to fix a government problem, what do you ask on Monday morning to avoid your own 10-year journey?

The problem you think is the problem is rarely the actual problem

Government doesn’t experience problems the way traditional businesses do. Understanding why leadership may not have full visibility into what’s broken is the first step toward engineering teams fixing causes when they’re handed symptoms. Let’s illustrate by walking through a common government problem-solution cycle.

Consultants lead an external audit and flag a data system outside the standard enterprise environment. This system, which we’ll call “rogue shadow IT” (RSIT), is labeled high danger because it duplicates the enterprise platform – this might mean that too many people have access to confidential data, that the process for using data is unsafe, and reports somehow show different outcomes. The obvious solution is to migrate the data and shut down the RSIT.

The internal team acts quickly, shutting down the RSIT, and the problem is solved. Or is it? Six months later, the same system is back and stronger than ever. Why?

First, a deeper look might have revealed that the RSIT was never, in-fact, rogue. IT built it and kept it running. It existed outside the enterprise platform because the enterprise platform couldn’t seem to do what the team needed. The external consultant didn’t get quite far enough into the weeds to understand that the RSIT was in fact a monitored, secure system.

Second, nobody had ever defined what “safe” data usage meant. Each new user, in a well-meaning attempt to protect confidential data, expanded that definition. The strictest reading of every rule and regulation won by default. (Marina would recognize this. An undefined standard always gets read in its most rigid form.)

Finally, you get to why the separate system is needed in the first place. Turns out, all this complexity exists to produce just one key report. So, the solution seems simple: scope and formalize a custom support within the enterprise system.

But building that report inside the enterprise system is hard because the data it relies on is inconsistent. The data is inconsistent because the data intake process was defined 15 to 20 years ago. People use the system the same way the very first user did all those years ago. The technology and needs changed while the process didn’t.

Nobody did anything wrong. Every choice made sense when someone made it. The system on its own was fine. The problem lives in how everything came together (or didn’t) over time, and how there was never quite enough incentive to adapt to an ever-changing world.

What you need is the toolkit to identify each of those small decisions and find and action their quick fixes, all while simultaneously keeping in mind the overall vision and the potential for more efficient redesigns.

Eight steps to find causes, not symptoms

1. Scope: who asked, and what do they actually control? Write down the problem you were given and who owns it. Then write down what that owner can change. Each time you find a process step out of their control, talk to the person who runs that step. You might find the world is much wider than you first expected.

2. Observe: who gave up, and why? Talk to users, especially the ones who quit. Their pain points will reveal what is process, what is systems, and what is people. At the VA, USDS research found over 70 percent of visitors had trouble with the health care application due to a requirement of using specific, outdated software to access it.

3. Quantify: where does the work pile up? As you map the process, count. Volume, wait time, and drop-off at each step. For Marina, a simple spreadsheet showed Rhode Island that over 80% of applicants never finished.

4. Trace: where does each requirement come from? For every step, ask whether it’s statute, regulation, guidance, a form, or habit. Then go read the source. For years, federal teams believed the Paperwork Reduction Act banned usability testing. It didn’t. OMB clarified this in 2016, then had to say it again in 2024 because the confusion was still blocking testing.

5. Compare: who does this well already? Find the peer who already solved it. Figure out how. Benchmarks are key to defining what good looks like. Marina put two states on a call to explain how each did things the other thought was illegal.

6. Constrain: if we fixed only this, would the outcome change? Your job is to find the binding constraint holding everything up. If the answer to the outcome change question is no, you haven’t found it. Keep going. Sometimes there are a series of binding constraints. This is great, because you can knock them down sequentially. Sometimes you have to handle constraints in parallel. This is tough, but that’s why you’re here.

7. Lever: are you fixing one output, or the machine that makes outputs? Fixing one step helps one pain point. Fixing the model solves the problem. California once ran three different online SNAP applications with 100+ screens of repetitive questions. With help from Code for America, California now has BenefitsCal, a single statewide system.

8. Teamwork: what’s the first (or second, or third) real quick win for everyone? Find the unglamorous solution or invest in the imperfect but workable fix that builds trust and gets you access. Too many builders make the mistake of prioritizing efficiency over empathy and perfection over good. Most problems have core human elements that need to be addressed before the system itself is fixed.

AI will speed up whichever way you’re going

How we deliver matters so much more now that AI is changing just how fast we can go down the rabbit hole.

AI can make the cascade of symptoms worse. A model reading a broken process will take it just as literally as any human and build off it. Maybe it will help you identify duplicative work or an out of order step, but since LLM’s tend towards sycophancy and work within existing systems, it’s (probably) not going to tell you to kill or reorder steps entirely - at least, not without the right prompting.

AI can also make tracing cheap and easy. Marina spent years comparing state rules by hand, one monthly call at a time. Today a practitioner could pull every state’s licensing code and ask where each requirement came from in an afternoon.

The key here is that a tool doesn’t pick the direction. The questions you ask to understand the problem before you use the tool are how you give it the direction it needs. Intentional use of AI can assist and accelerate the eight steps above, driving work toward long-lasting fixes instead of temporary band aids.

How to tell you’re still peeling

The hardest part is knowing when you’ve hit bottom. A few signs you haven’t:

• The fix helps one site, project, or process – but not the next one.

• The fix adds a rule.

• Someone says “that’s the law” but can’t point to the exact line.

• The request is for a report or dashboard.

• You write articles to convince folks to join government and tackle problems with you

Are you a government practitioner who wants to use AI and technology to do your job more effectively and efficiently? A technologist who wants to dip your toe into the civic realm? An expert in government who thought our article was spot-on, completely off-base, or somewhere in the middle? Tell us more by filling out this short form!

Clearly, AI is a powerful tool – and governments have the opportunity to use AI to address longstanding problems. But they must also maintain commitments to safety, fairness, and accountability to the public. Achieving both goals means increasing adoption of AI while monitoring its use and ensuring that people remain accountable for government actions and decisions.

In the near future, AI will help deliver social safety net programs to meet people’s basic needs, set up pathways for people to get well-paying and fulfilling jobs, and even build homes and infrastructure. As a general purpose tool, it will speed up day-to-day work that involves analyzing, reasoning about, and communicating information in operations, management, and rulemaking and legislation.

The keys to making AI work for humans, and not for corporations, are the same government oversight measures that exist already, streamlined and enhanced to ensure oversight over AI as well.

Here’s how this would work.

Efficient operations

A standard application for SNAP must be processed by a state agency within 30 days. In practice, the average state processes only around 85% of applications on time, meaning millions of people spend weeks or months waiting for money to buy food. AI tools can help caseworkers more quickly answer questions from applicants and scan submitted documents for common issues; similar use of AI in the private sector allowed customer service staff to process requests up to 34% more quickly. This can speed up case processing times by shortening the time it takes to complete routine steps and letting applicants know more quickly if they need to provide more information to the agency.

Agencies can also implement oversight measures to ensure continued accuracy. They can randomly sample some tasks to be completed only by humans to ensure that automated systems behave in ways equivalent to people. They can have the systems flag complex cases and inconclusive results for extra review. With efficiency gains from AI, human caseworkers can spend more time listening to applicants to ensure that a person’s real-life situation is properly represented by a technical system.

Agencies are ultimately accountable to external oversight; for SNAP, these include fair hearing protections where applicants can dispute the results they received and independent quality control assessments by state and federal governments. Agencies routinely make operational changes unrelated to AI; these oversight measures ensure that any operational change doesn’t come at the expense of accuracy.

Data-driven management decisions

What can governments do to set people up for financial success? The evidence for what works is limited, and the effectiveness of any employment program will vary from person to person and across communities. Researchers have used data on millions of people in recent years to shed new insight on what has driven change across the country. But state and local government agencies tasked with improving employment want to know not only what works overall but what can work for them.

AI can help program administrators make more informed decisions, by analyzing local data (including novel data sources like social media) and searching for comparison cases to see what worked (and didn’t) in cities and states like theirs across the country. Administrators can treat these like scaled-up case studies, providing them with ideas for measures to implement and evidence for which interventions have been particularly robust.

Program administrators rarely defer judgment to case studies, though, and should not defer judgment to AI either. They can continue to draw on their practical expertise to sanity check AI analyses and make changes to program rules and funding allocations based on what they know to be feasible.

Laws and rules that actually work

Builders in New York City must navigate 40 approval processes to get something built. Other construction and infrastructure projects like for renewable energy also face onerous permitting processes. Local, state, and federal governments around the country are pursuing permitting reforms so that homes can be built where people want to live and clean energy can be generated to meet growing demand and transition away from fossil fuels.

The high volume of laws and regulations, especially across different levels of government, mean that even listing out all permitting rules that builders have to comply with is a challenge. AI tools can help gather that information so that legislators can ensure reform is comprehensive.

Legislators are often pressed for time and drawn between competing priorities. Even before AI, they would often rely on outside groups to provide the text for legislation, and often weren’t given enough time to fully read the bills they voted into law. AI has the potential to magnify these problems; it is already common for legislators to use AI to draft laws. Legislators must ensure that the bills they draft and vote on receive as much human review and scrutiny as laws created before AI.

Government must rise to the occasion

Unlike with math, government is built around human judgment and accountability. Success in government doesn’t always have deterministic, provable outcomes. Deciding how to harness AI while maintaining safety and accuracy standards is a major open problem, and even top tech companies are figuring out how to get it right. The responsibilities, accountability, and governance of a public agency are different from a private company, so the best practices for how to use AI in government will also differ in important ways.

It is therefore imperative that governments take the lead in responsible AI deployment instead of letting the private sector decide for them.

A generation ago, computing technologies like email, spreadsheets, and the internet fundamentally changed how office work was done. Employees’ responsibilities changed, and government agencies had to deliver on more because the public expected more. The same thing is happening with AI today.

We’ve invented robots that can perform mathematical miracles. Governments can harness these robots responsibly to ensure that people can put food on the table, find well-paying and fulfilling jobs, and have homes to live in and electricity to power them. A government that fully delivers on what the public wants is its own miracle.

Are you a government practitioner who wants to use AI and technology to do your job more effectively and efficiently? A technologist who wants to dip your toe into the civic realm? An expert in government who thought our article was spot-on, completely off-base, or somewhere in the middle? Tell us more by filling out this short form!

• What: Public Engagement Hackathon, our second flagship build event.

• When: Wednesday, September 30, 5:30pm to 10:30pm. Matchmaking at 5:30, builds from 6:15 to 8:30, demos at 8:30, social hour after.

• Where: Sign up to view location

• Sign up: Register on Luma. Dinner is on Civic Roundtable.

• Partners: CUNY PIT Lab, Civic Roundtable

• Judges: Raj Korpan (CUNY PIT Lab), Jeremy Ney (American Inequality), Sara Lind (Open Plans), Amaury Dujardin with more to be announced.

• Want to find teammates before the event? Be the first to hear about future events? Help build the State Capacity Ecosystem? Join our Slack.

• Keen to find more Ecosystem resources? Check out our new website.

At our first hackathon, 80+ attendees produced 20 working tools that addressed real government tech gaps in under two hours. The takeaway wasn’t just the technology; the magic was in watching the force multipliers of the State Capacity Ecosystem. When a government practitioner who knows a problem sits next to a technologist who can build against it, AI becomes a tool for inclusive building and unlocks the ability to construct useful working prototypes on the spot.

On September 30, we’re running it back. This time, the night’s problem statements are centered on how New York actually hears from New Yorkers. The city has no shortage of input channels: it has 59 community boards, a 311 line, participatory budgeting, and public hearings on everything from liquor licenses to land use, all meant to allow public feedback into city processes. But, in many cases, it lacks the infrastructure to make the information generated from these channels actionable, representative, equitable, and connected to outcomes.

Stream One: Community Boards Ops
_{CBs have a lot to listen to and track, with few supporting tools}

Community boards (CBs) are the city’s ground floor for public input: every land use action, liquor license, and street redesign passes through one of the 59 boards, and each board runs on volunteer members, a staff of two or three, and whatever filing system it has improvised on its own. Unfortunately, tech tooling for CBs is nearly nonexistent (and probably not a historically attractive venture capital market).

Here are three possible build paths for participants:

• A meeting assistant. Robert’s Rules and parliamentary procedure trip up new members constantly. CB resolutions get drafted from scratch each time. A procedural helper or a resolution writing agent might solve hours of headache for CB members.

• A member educator. Term limits from the 2018 charter revision force the first wave of open community board seats in 2027, which means hundreds of new members learning the job at once, and onboarding today is weak. An AI assistant or a data-backed member guide would arrive at exactly the right time.

• An end-to-end resolution tracker. Boards pass resolutions addressed to agencies: Transportation Committee resolutions to the Department of Transportation (DOT), Parks Committee resolutions to the Parks Department, so on. Then, in many cases, nothing visible happens. A tracker that connects each resolution to its target agency and follows what the agency actually does would give boards clear evidence of how the government listens and acts.

The data is plentiful but dispersed. NYC Planning’s Zoning Application Portal carries every Uniform Land Use Review Procedure (ULURP) action, the Community Boards dataset is on NYC Open Data, and the NYC Council’s Legistar has a public API. SCE’s own Tal Roded has already aggregated all 59 board websites, members, and leadership in one place. And we should plug Tal’s Block Party, which has meeting transcripts and extracted resolutions ready to query for many community boards.

Stream Two: Public Voice At Scale
_{Governments take in a lot of citizen feedback, but is that feedback inclusive, fair, and effectively digested?}

City Hall frequently holds hearings to understand public sentiment on high priority topics and issues. Agencies engage with communities to learn how they can best help constituents. New Yorkers filed 3,655,040 requests through 311 in 2025, about 10,000 every day. Democratic participation is great, but how and when citizens interact with the government can introduce significant bias and complexity to analysis: who gets to offer the feedback, how is each piece of feedback weighted, and who hears it? Ideas here include:

• A complaint-pattern dashboard with a residual map. Model expected complaint volume for each district from its conditions, then map the gap between expected and actual. Where complaints fall well below expectation, you have likely found under-served blocks. Experts like Vital City’s rolling “CrowdVox” interviews could then validate whether the lack of complaints might mean community satisfaction or resignation.

• A citizen-observation toolkit. The Central Park Squirrel Census proved that volunteer observers can produce a real dataset (3,023 squirrels, counted block by block). How might the city harness citizen energy and collective observations for other needs: on potholes, blocked bike lanes, broken park lighting, or otherwise? What tools might facilitate the collection, analysis, and display of such data?

• Tools that serve the marginalized voice. Cluster participatory budgeting ideas by neighborhood and theme, mine the Council’s hearing archive to analyze public testimony at scale, or geo-tag social media conversation to neighborhoods, topics, and agencies, turning real unstructured public voice data into something an agency or a board can act on. Build agents that support language translation where human translators might not be available. Leverage technology to reach folks who the government does not reach today.

Participants might utilize the 311 Service Requests dataset, which covers every request since 2010 with location, type, agency, and resolution fields, as their base data layer. Social media data or public comment data are under-used by the city and provide live feeds of the public voice. Or builders might instead prioritize tools that help generate new data from historically underserved and underheard constituents.

Stream Three: Judge Specials
_{Our Judges are mulling over problem statements of their own. Play it by ear and generate novel solutions day of.}

This is a dealer’s choice category. Here are a couple ideas our judges have floated thus far, and we expect more to follow:

1. How can we create a Big Tent that includes many different voices and organizations but does not find itself struggling with an ‘everything bagel liberalism’ where nothing gets done because too much has been bitten off?

2. Could we map the areas around transit stations in NYC that are not zoned for residential but probably should be to increase transit-oriented housing (these would be C8 or M1 zones)? With a particular focus on the Bronx.

Get to know your judges

Our judges are experts in these topics, and they’re excited to see what you can build for them.

Raj Korpan, CUNY PIT Lab.

Raj Korpan is an Assistant Professor of Computer Science at Hunter College, CUNY. Raj researches artificial intelligence, human-robot interaction, explainable AI, and socially aware robotics through Hunter’s Trustworthy, Intelligent, and Explainable Robotics (TIER) Lab. Expect him to push teams on who their AI is designed for, whose perspectives are missing, and whether they have built meaningful mechanisms for trust, explainability, inclusion, and community participation into the product.

Jeremy Ney, author of American Inequality.

Jeremy maps America’s regional divides county by county for more than 25,000 readers, work that has been covered by TIME, The New York Times, and NPR. He is also a full-time employee at Google and teaches as an adjunct professor at Columbia. Expect him to push teams on how data can reveal the less obvious relationships behind social problems, distinguish signal from conventional wisdom, and turn those insights into solutions that can meaningfully improve people’s lives.

Sara Lind, Co-Executive Director at Open Plans.

Sara Lind is Co-Executive Director at Open Plans, a nonprofit with the mission to transform how people experience New York City’s streets and public spaces. She has guided major public policies, driven agendas in City Hall and in Albany, and shaped the narrative on how to make New York’s streets more livable, safer, and joyful. Expect her to push teams on how technology can create stronger connections between public voice and the public realm and give New Yorkers a meaningful role in shaping the streets and spaces we share.

Amaury Dujardin, Public Policy Manager at Citizens Union

Amaury is a Public Policy Manager at Citizens Union, a nonpartisan good-government organization working to make New York’s government more open, effective, and accountable. He helps lead legislative advocacy in Albany and City Hall and manages the organization’s candidate evaluations, drawing on nearly a decade in politics before turning to policy. Expect him to push teams on how technology and data can close the gap between constituents and the officials who represent them, so that reaching out to an elected official feels less like shouting into a void and more like actually being heard.

Additional Judge to be announced!

How the evening runs

• 5:30 to 6: Facilitated team matchmaking. Color-coded name tags mark your problem-area interest and role so you can find a team quickly. You can also join our Slack to find a team ahead of time.

• 6 to 6:15: Introductions. A short welcome from us, partners, and judges.

• 6:15 to 6:30: Breakout Q&A with the judges, then final team selection.

• 6:30 to 8:30: Build time. Dinner arrives around 7, and experts run hourly check-ins.

• 8:30 to 9:30: Demos. Three minutes per team: one minute of demo, one minute of judge Q&A, and a bit of buffer because one minute is never enough.

• 9:30 to 10:30: Informal social hour at Clinton Hall (16 W 36th St., New York, NY 10018)

What winning teams get: Big kudos, a spotlight feature on our website and at a follow-on demo night, and maybe even a special surprise on the day-of.

Come build with us

Our June event proved the State Capacity Ecosystem is real and that this community can ship viable solutions to important problems. September 30 asks builders to use those skills to help New Yorkers feel heard. If you work in or around government, study it, advocate for a neighborhood, or build tech products, there is a team on the 30th that needs exactly what you know.

Register for the Public Engagement Hackathon on Luma

While you are at it, join our Slack to follow the run-up, subscribe to this Substack to stay in touch, and check back in on this post for future updates.

Leave a comment

Share

But while the government builds muscle for yesterday’s digital problems, the private sector is already solving the next challenge: deploying AI. Sit this one out and the government risks falling into the same patterns that got us here in the first place.

Subscribe now

The rise of the Forward Deployed Engineer

AI can be a transformative technology. From data extraction to process completion to rapid prototyping, tools available today can increase productivity by multiples.

Applying AI well is difficult. It means mapping a process in detail, understanding why it works the way it does, and knowing exactly where AI helps and where it doesn’t. This work requires technically savvy individuals capable of embedding themselves inside client organizations, building trust, and delivering value. Tech calls this role the Forward Deployed Engineer.

Palantir built an entire business on embedding engineers inside client organizations to ensure their technology worked in production. Every major AI company has since copied it. Job postings for the FDE role are up over 700% year over year, and a July census of just eleven companies found close to 300 open roles. OpenAI launched a $4 billion subsidiary built entirely around the role in May. AWS committed another $1 billion in June. S&P Global found that 42% of companies abandoned most of their AI initiatives in 2025, up from 17% the year before, proof that deployment expertise, not the technology itself, is where the real value lies.

Acknowledging that properly deploying AI seems to require these talented individuals, the question becomes whether to rely on vendor expertise or build in-house. This is both an equity question and an operational one.

Forget FDEs: government must hire its own AI engineers

An AI system built by a vendor optimizes for two things: contract compliance and profit. An AI system built by a government AI engineer can optimize for the resident calling about a benefit, the small business owner filing a permit, the parent who needs a form in another language.

Done right, in-house AI deployment makes government fairer and more inclusive. AI tools can be designed to meet residents where they are, address a laundry list of edge cases that would otherwise go unsolved, and allow for live prototyping, feedback, and iteration. Left to a vendor optimizing for contract terms and resale to the next twelve agencies, the same tools tend to default to whatever is contractually aligned, not what actually serves the public. The constituents most in need might not be able to use the tool at all.

The operational case is just as strong. Policy does not sit still. Tal Roded recently mapped how many standing obligations New York City agencies answer to at once: more than 8,000 obligations from the City Council, roughly two-thirds of them recurring every year, with new ones added constantly. Every one of those can change what an agency’s AI workflow needs to do next month.

A tool only the vendor can change is a tool the government does not actually own and cannot actually use. A tool agency staff can adjust themselves is the difference between fixing a workflow in a week and renegotiating a contract for six months. Just like with other technology, vendor-built AI bills you every year for the privilege of staying dependent.

What building AI in-house looks like

New Jersey’s Office of Innovation is already modeling in-house AI expertise. The state was an early adopter of AI through the NJ AI Assistant, an in-house chat tool that allows thousands of State employees to work more efficiently, hosted on secure infrastructure at roughly 1/20th the cost of a commercial product. The office has since built reusable tools for agencies to speed up tasks like reviewing documents and drafting memos. These tools turned weeks-long reviews into hours of work, while keeping humans in the loop and honoring the high standard of trust residents expect of the government.

Congress is also weighing how to actualize its own AI adoption. Representative Stephanie Bice, who chairs the House’s Subcommittee on Modernization and Innovation, requested funding this summer for Mia, a secure AI agent connected to data across the House, built through a pipeline that trains staff to build these tools themselves instead of buying them off the shelf.

Delivery ultimately reflects the same model State Capacity advocates have championed for years. A government AI engineer sits with staff members before writing any code, long enough to learn what the decision actually requires. They curate the evaluation set: what a correct answer looks like, on which cases, at what error rate, who reviews the misses. They own the data pipeline, so accounting for a new team process does not require a full rebuild. They outsource when necessary, but they read the contract closely enough to know what is being given away and how to hold the vendor to account. And they build the small thing in-house when the small thing is enough, which is more often than a vendor pitch will admit.

As for the cost of that engineer? Agencies already pay for this work, just buried inside an implementation line item instead of a headcount line. New York’s entire PIT Crew initiative costs less than what a large agency spends on a single modernization contract.

Digital was never the point

Colorado’s real lesson is that the government needs to and is capable of responding quickly and authoritatively to emerging technology. AI is just the next evolution of that challenge.

We cannot afford to wait. Talent is being snapped up by AI labs, start-ups, and consultancies. Governments own very little of the AI expertise today, and every month that passes, more of it accumulates elsewhere.

The agencies that hire their own AI engineers now are the ones who get to decide how AI actually shows up in their residents’ lives.

Are you a government practitioner who wants to use AI and technology to do your job more effectively and efficiently? A technologist who wants to dip your toe into the civic realm? An expert in government who thought our article was spot-on, completely off-base, or somewhere in the middle? Tell us more by filling out this short form.

An influx of digital talent is a blessing for state capacity. Not every problem has a tech solution, but in many cases a well-built website, software tool, or data platform makes a solution far more tractable. There is a reason the nation’s first Click to Cancel protections shipped with a PIT crew tasked with implementing them.

But if we want to see these efforts be successful, the Mayor’s office should go further and fully embrace responsible and trustworthy use of AI across city government. By doing so, it will ensure it can meaningfully deliver on campaign promises: access to cheaper groceries, building more affordable housing, and making it easier to build a thriving small business in New York.

Subscribe now

Trust & Capacity Building

Getting this technology right is central to rebuilding public trust, which now sits near the lowest levels on record (just 17% of Americans say they trust the government to do what is right). That’s the bet the mayor is making: if you deliver services for people, they’ll trust you.

New Yorkers reach that judgment through the local transactions that either work or don’t: renewing a permit, enrolling a kid, getting a benefit they qualify for, reaching a caseworker before a deadline. But to rebuild this trust at scale, it’s going to take a whole lot of government effort, and 30 technologists can’t fix how a 300,000-person workforce serves the city. To do that, the government has to work smarter, not just harder.

The tools to work smarter finally exist. Government runs on unstructured text: applications, licenses, benefit rules, bookkeeping, inspection records. That is exactly the work AI made cheap almost overnight. At a two-hour hackathon we ran last month, teams built working prototypes that agencies had in some cases spent years developing. For the first time in a while, maybe since the invention of the internet, the gap to build is as small for an agency as it is for a business.

If the city government wants to take full advantage of this shift, it should act quickly and at scale. That means making it easy to solve many of the mayor’s top priorities with this technology. It also means equipping more people in public service with AI subscriptions and the training to build live solutions themselves (at the time of writing, many of the city’s agencies still have no work-approved LLM tools, including for coding and development).

It is precisely because AI is powerful that the government should be proactive about its equitable, fair, and just use. Avoiding the conversation risks repeating old patterns where agencies get locked into private vendors and end up renting systems they cannot change. Real change happens the other way: by putting the technology in the hands of practitioners and letting them deliver.

Enough Talk – 3 prototypes of AI workflows for Mayoral Priorities

We would rather show than tell. What follows are prototypes built in hours, each for a few dollars in token spend, against three of the mayor’s stated priorities. Many more could easily follow.

Making Groceries Cheaper

Food is one of the fastest-rising costs for working New Yorkers, and it hits hardest in the neighborhoods with the fewest options. Mayor Mamdani’s pilot of city-owned grocery stores is meant to bring down prices. But where to position the five stores? How to define what ‘cheap groceries’ means? How to publicize outcomes and hold vendors to account?

We identified disadvantaged census tracts suffering from food insecurity with transit connectivity and used public data sets to build an interactive dashboard.

Expanding Ownership And Making Small Business Easier

Worker cooperatives and small businesses build stable, local wealth while offering real worker say. But they are hard to start and run. How can the Mayor lessen upfront costs associated with incorporating a business? What tools might help with navigating licenses, inspections, and fines across dozens of agencies?

We aggregated city regulations and built a simple guide for cooperatives and businesses on what to do after receiving a summons, addressing an issue mentioned in the Mayor’s OPEN for Small Business plan.

Building affordable homes

Mayor Mamdani is aiming for 200,000 new affordable homes to be built as part of his Block by Block plan. But builders currently have to navigate 40 approval processes across 18 offices and agencies, according to his COGE charter revision committee.

We used real public data and real-world examples of projects to build new homes and illustrated the pathway from idea to official approval based on characteristics of the buildings.

Takeaway

Mayor Mamdani has a rare opportunity to use public-interest AI to make our city more affordable. AI is already reshaping how people work. For those of us who want the government to truly serve the people, this inflection point is an opportunity: a chance to catch up where government has lagged and own how and where technology is used.

It is time to lean into all the amazing possibilities that AI unlocks and create our own vision for what good government means.

Leave a comment

Share

It turns out one of our hackathon Judges built the underlying infrastructure to make this possible, and a similar dashboard. It took years. The hackathon team built their prototype in two hours.

In all, 80+ people split into 20 teams and built 20 real tools. Check out this clever video game that teaches New Yorkers how to use 311. Browse a website indexing NYC’s civic data. Fill out a sample benefits form simply by speaking. Use this tool for searching and reviewing city procurement records. And see here for more.

Our takeaway: AI offers a path to more inclusive building

AI has democratized the ability to produce. Anyone can now theorize, scope, and build something new in a matter of hours. AI is an imperfect tool and comes with real considerations about environmental impact, security, and responsible usage. It is not the right tool for every task, nor should it exist unregulated. But as those conversations continue, AI’s impact ripples through society.

On Wednesday, we saw the best of that. Our hackathon was not just about faster builds. AI put the ability to create new technology in the hands of public sector experts who previously did not have that power.

Technologists still built, but government workers informed, flagging when the tool didn’t match real workflows and pushing for features that would actually get used. Researchers identified key data sources and ran live evaluations. Advocates reviewed UX and ensured design accessibility. Feedback was live and specs changed with each iteration.

One of our practitioners said it best:

“AI meant for the first time we could weigh in on design and product decisions instead of beelining towards writing code.”

AI done right is a tool for access, equity, and self-governance

Consider the above. Now imagine an institutionalized version of “hackathons” that bring folks into the building process. Community engagement, perhaps at events physically situated in relevant communities, where practitioners build prototypes for their constituents and receive immediate feedback. This means real people contributing to end products they will actually use, and the ability for voters to literally see changes being made in minutes, not years.

Contrast this with the standard government technology approach that involves a long requirements document, an outsourced legacy vendor, a long waiting period, and a product that is only 70% right and locked in by contract.

Wednesday night showed how AI can take people previously outside the conversation and put them in the driver’s seat.

Government has a tendency to wait; it doesn’t have time

Government has always moved carefully, and for good reason. Delaying judgment on AI feels careful. But careful and slow are not the same thing, and right now slow is dangerous.

Our hackathon proved AI tools can now query live government data — 311 records, permit filings, property data, contract databases — and do something real with it.

Researchers at MIT are already building the infrastructure for these tools to discover each other, share data, and work together across systems. Project NANDA is an open-source effort to build what they call an internet of AI agents: a decentralized layer where software tools can find each other, verify credentials, and coordinate tasks. Anthropic, OpenAI, Google, Microsoft, AWS, and Cloudflare have already agreed on open protocols for these tools to communicate across platforms. NIST launched a formal standards initiative for this infrastructure in February 2026.

AI infrastructure is being built now and government data sits at the center of it. The question is whether government has a role in shaping how that goes or if it finds out after independent actors decide.

Government needs to give itself the right tools

Procurement in government is often built for risk avoidance, not outcomes. Vendors with less presence but better technology are sidelined in favor of ‘more feasible’ deployments. But reaping the benefits of emerging technology often requires embracing emerging vendors and meeting public servants where the work happens.

Wednesday made it clear that AI’s power to democratize creation relies on its most cutting-edge features and functionality. The biggest opportunity is not just giving government access to AI, but embedding it into workflows, systems, and partnerships.

Leading AI companies offer a differentiated set of tools compared to legacy vendors who have adapted to AI. The practitioners who showed up last week know the difference. They are not asking for a bolt-on that helps with email drafting and faster search. They are asking for the ability to build.

Government is already moving

California announced yesterday it is rolling out Anthropic to state agencies.

But inclusive AI requires more than access to the latest models. The transformation means new ways of piloting, shipping products, and evaluating success.

Pennsylvania has over 3,000 state employees using ChatGPT Enterprise, with thousands more in training. They rebuilt unemployment benefits chatbots to handle customer interactions and are using AI to flag document errors for benefits applicants, catching problems before they cause delays.

New Jersey became the first state in the nation to enshrine an innovation office into law, making it permanent across administrations. Their team built an AI assistant for state employees at $1 per user per month versus $20 for off-the-shelf commercial tools. Jennifer Pahlka, one of the most prominent voices on government reform, has publicly called New Jersey a model for responsible AI use in government.

Maryland, after building to more than 50 agencies actively using AI tools, recently launched an AI Innovation Lab to grow pilots into long-term products embedded in state infrastructure.

What these states did was not complicated: they hired teams, gave those teams real tools and real problems, and let them build. It just took vision and decisiveness.

We’re growing a community of practitioners who want to build

We’ll keep hosting events bringing together practitioners who want to help government increase its own capacity.

• Large hackathons that show what this community is capable of

• Targeted hackathons working directly with a single agency on a specific problem to drive real impact

• Demo nights that help the output get in front of decision-maker

• Speaker and salon-style discussion nights connecting real experience to desire for further action

• Expanding the State Capacity Ecosystem Hub to make collaboration easier

Our NYC community is a model. Our tools and tactics are open source (or will be, in forthcoming posts). Inside NYC or out, if you want to get involved, reach out: we will help you build community wherever you are.

And look out for more soon.

Thank you

A hearty thank you to everyone who attended the hackathon. Thank you to our partners at Civic Roundtable and CUNY PIT Lab and to our judges Dan Steinberg, Josh Seiden, Steph Yim, and Ben Kallos for their participation. You were the ones who made this all possible.

If you, the reader, are seeking to get involved more directly, several job opportunities were highlighted at the event:

• Dan Steinberg is recruiting technologists to work with him at the new NYS Office of Performance Management and Innovation (OPMI). Contact him for more information at daniel.steinberg@exec.ny.gov.

• NYC Office of Technology and Innovation (OTI) has a number of job postings for product managers, engineers, and product designers. Apply here.

• Civic Roundtable is hiring for a Chief of Staff and will have a few more roles opening soon for people excited to help public sector teams collaborate, share knowledge, and build capacity. See open roles here.

Leave a comment

Share

Please note that the views expressed in this article are not meant to represent the official position of any associated individual, organization, or employer.

TL;DR

We’re hosting an AI Hackathon on June 24 in New York City, 5:30 to 9:30 pm. Attendees will pick a real problem, join a team, and ship something by the end of the night.

Within a week of our launch, we hit the registration cap for the event. For those who did not register but are interested in getting involved, we recommend subscribing to this Substack and adding yourself to our Connect page. Future hackathons and events will be announced here first.

Thank you to our partners, CUNY PIT Lab and Civic Roundtable.

Read on to learn more about the event, including our announcement of the project tracks, judges, and logistics.

Why the hackathon exists

Our build night is a forcing function. Real problems. People from different corners of the State Capacity ecosystem. Grassroots community, hacky approaches. The point isn’t a finished product by 9:30pm. The point is to start putting practitioners and technologists in the same room, on the same problem, at the same time.

You don’t need to be a developer. You only need to care about the problem and be ready to experiment.

How the night will run

5:30

Doors open. Pick your track, find your team, grab a slice of pizza.

6:00

Kickoff: welcome, format walkthrough, and a brief from Civic Roundtable on building for the civic sector.

6:15

Judges lightning round: 2–3 judges share one problem they’ve seen up close and what they want to see built in their track.

6:30

Teams form and build. Judges and staff roam the room.

8:45

Demos begin. Three minutes per team: a working prototype, dashboard, or AI-assisted workflow. No decks.

9:10

Judges deliberate and name standout teams.

9:25

Wrap: next steps, community announcements, and how to stay connected.

The tracks

Projects are suggested to follow one of three tracks – Mayoral Priorities, SNAP and Benefits Delivery, NY Open Data – or can be submitted under the Open category.

Each track has a real problem statement sourced from a practitioner.

Mayoral Priorities: Agents supporting the incoming administration: grocery store siting, permitting reform, tools to help small businesses navigate city services.

SNAP and Benefits Delivery: Eligibility prescreening, application support, and caseworker tools built for a policy environment that keeps changing.

NY Open Data: Making NYC’s open data actually usable: plain-language Q&A, consultation summaries, dataset-to-answer pipelines for people without a data team.

Open: Bring your own live government problem and recruit collaborators on the night.

The judges

Judges and supporting staff from New York Government, Academia, GovTech (and more) will roam the room, give feedback, and name standout teams at the end of the night.

Dan Steinberg

Dan has spent over a decade building the data and performance infrastructure that makes city government actually run. He’s now leading NYS OPMI: an office recently created by Governor Hochul to strengthen statewide performance management systems, support data-driven policymaking, and help execute cross-agency initiatives. He knows what operationalizing technology inside the government looks like.

• Executive Director of the new NYS Office of Performance Management and Innovation (OPMI)

• Director, NYC Mayor’s Office of Operations

• Winner, NYC Hayes Innovation Prize for IT Leadership

• Built the Dynamic Mayor’s Management Report: real-time performance monitoring across 46 city agencies and ~22,100 performance indicators

Josh Seiden

Josh co-founded Civic Roundtable and has spent the past several years in the room where the actual work happens: onboarding agencies, running trainings, making sure the technology lands. He’s a co-host of this event and brings a ground-level view of what it takes to deploy tech inside government.

• Co-Founder, Civic Roundtable — collaboration platform used by public servants at 1,500+ agencies across all 50 states

• Former Senior Director of Strategic Initiatives, Mark43

• Former Human Capital Consultant, Deloitte; focused on systems implementations at the Department of Homeland Security

Stephanie Yim

Steph Yim is an experienced multi-disciplinary designer and strategist leading design, research, and implementation of public services at the intersection of government and civic tech. She is a practitioner and educator of community-centered design. As the design operations director at Public Policy Lab, Stephanie partners with government agencies and non-profit organizations to design improved ways to deliver public services to at-risk communities. Steph is a proud native of Queens, NY and daughter of immigrant parents.

• Worked for agencies like the United States Digital Service, NYC Department of Education, Robin Hood Foundation, and the NYC Mayor’s Office of Economic Opportunity

• Co-founded Good Call, NYC’s first 24/7 arrest support hotline

• Over a decade of experience in UX and product design in education, housing, childcare, and benefits access

Benjamin Kallos

Ben has spent his career using technology to make government more open, accountable, and easier to navigate, first as a self-described “political geek squad,” then across all three branches of government. As a New York City Council Member he authored some 60 laws, and most recently he worked in the White House leading cross-agency data-sharing initiatives to make it easier for residents to access human-service benefits. He knows what it takes to turn civic-tech ideas into law and into working systems.

• Former New York City Council Member (2014–2021), representing the Upper East Side, Midtown East, Roosevelt Island, and East Harlem

• Former member of the U.S. Digital Service in the White House, leading cross-agency data sharing to streamline benefits access

• As Chair of the Governmental Operations Committee, worked to root out waste and use technology to improve access to government, including putting NYC’s budget online

• Authored the full public matching campaign-finance system and the affordable-housing law behind Housing Connect (housingconnect.nyc.gov)

• Attorney at Law, founder of Kallos Law, representing communities, candidates, organizers, and small businesses

What’s next

This is the first version of something we want to keep building. After June 24, we’ll publish a write-up: what teams built, what judges noticed across tracks, where the gaps were, and commitments for what’s next. Subscribe to the Substack if you want to see it when it drops.

And if you want to stay connected to the broader ecosystem — the orgs, the tools, the funders, the fellowships — the State Capacity Ecosystem Hub is here: State Capacity Ecosystem — NYCuriosity Data.

Leave a comment

Share

The State Capacity Ecosystem Hub is live.

• Search (in Beta) curates a personalized view of who’s working on your specific areas of interest

• Directory lists >300 orgs focused on government effectiveness and what they’re working on

• Connect lets you add what you’re working on, seek and share opportunities, and find collaborators

• Affinity helps you find surprise relationships between organizations

Two key things to do right now:

• Submit yourself to Connect — tell us what you’re working on and what you need; we’ll curate a living database of challenges, funding opportunities, and practitioner partners

• Sign up for our first event, a CivicTech AI Hackathon — June 24 5:30-9p

Credit to , my co-author on this post and builder of the Hub’s website.

Why we’re building the Ecosystem

Good policy fails because the practitioners, technologists, researchers, and funders working to fix government are operating in silos, without a clear view of each other and with few collaboration pathways. Great talent is left out of the conversation not by intent, but because it simply cannot find its way in.

We believe visibility is infrastructure. A clear map of who is doing what is the first step toward State Capacity acting like an ecosystem and not just a collection of isolated players.

Our goal is to help create that well-functioning ecosystem: one where every participant knows what the others are doing, sees them as collaborators rather than competitors, and can access shared knowledge and resources when they need them. Great theory gets operationalized. Hard-won practitioner knowledge travels. Capital flows toward the gaps, not just the familiar names.

Overview of the Tools

Search (Beta), located on the Hub, returns results relevant to you based on a mad libs-style interface that allows you to select your role, what you’re seeking, a problem area, and a geography. This feature can only show what currently exists and will improve as we add more organizations and individual connections.

Directory offers you a view of all the organizations in the Ecosystem. Use it to understand who is working on what. Filter based on segment, problem statements, geographic focus, and more. It offers a more holistic view versus Search’s curation.

Connect is the layer we’re most excited about. Find real people and organizations working on problems you care about. More importantly, submit what you’re working on so that others can find you. Together, we’ll curate a living, breathing picture of who’s working on what in the Ecosystem.

Affinity helps you find surprise relationships between organizations based on shared problem areas, overlapping funding sources, and description similarity. Use that knowledge to reenvision who you can work with and on what.

About the Event

On June 24, we’re hosting a one evening hackathon for civic technologists, government practitioners, students, and anyone else who wants to build tools that help strengthen our government’s ability to deliver.

The goal is connection, not perfection. Find collaborators interested in the same topics. Build quick, working prototypes that take ideas from -1 to 0 and show whether an idea is worth pursuing. See if someone in attendance can help connect you to the right folks to turn that idea into something real.

Check out the event page for more information.

How you can help

When we delivered the first version of the Ecosystem, we effectively shared a list of orgs; and an incomplete one at that. Folks appreciated the clarity it brought to who is working on State Capacity, but there was no way to act on that clarity. Dozens of follow-on conversations informed how to make our map more useful: living problem statements, the ability to find collaborators, and hosting live events move us from theory to action.

Chances are this version of the Ecosystem isn’t perfect either. You can keep helping us improve it. Add your organization and your problem statement. Join our events and host your own. Reach out with feedback.

Our hope is that our work echoes the lessons of State Capacity: agile delivery, test and learn methodologies, shared evidence, scalable results.

The closing message

Delivering State Capacity may be hard, but the roadblocks aren’t mysterious: they’re structural. Knowledge doesn’t travel because there’s no infrastructure for it to travel on. Capital concentrates because funders fund what they can see. Practitioners work in isolation because isolation is the default.

This is why we’ve built the Ecosystem, but it is only as helpful as the community behind it. And there’s a million other ways folks can build out the infrastructure further in their own ways.

The State Capacity Ecosystem already exists. It just needs to act like one. Let’s build it together.

For years, Jen Pahlka has written on how to improve Government’s ability to deliver. She uses State Capacity to define what good government looks like: hiring and firing the right people, updating old procedure, investing in technology, and instituting test and learn systems to drive outcomes.

I love her work. It has informed and inspired me. But operationalizing her theories has proved difficult (and Pahlka has delivered a massive step forward here with her new Recoding America Fund).

There are many thought pieces that detail why State Capacity is tough to deliver; very few help interested parties understand: what can I help with, how can I get involved, and who else is thinking about the same things?

My Contribution

There is a real ecosystem of people and organizations — some inside government, most outside it — already working on exactly the problems Pahlka describes. Founders building tools agencies actually need. Fellowships placing technologists where they’re missing. Researchers documenting and benchmarking results. Funders backing the work.

But this ecosystem has never been treated like one. My research revealed targeted resources, but no shared directory; no common language; no helpful landscape overview for newcomers.

The capital is there. The talent is there. The connective tissue is not.

So I built it.

The State Capacity Ecosystem v1 (225 orgs)

Civil Society

• Research (42) — Think tanks, university centers, and independent research institutions producing knowledge on government performance, digital governance, and state capacity.

• Fellowships (22) — Programs that place a person inside a government team, spanning early-career pipelines, mid-career rotations, and legislative placements.

• Community (21) — Volunteer civic tech groups, practitioner peer networks, and member-driven communities connecting people working at the intersection of technology and government.

• Advocacy (17) — Organizations pushing for government modernization, tech talent pipeline development, and stronger public interest technology policy.

• Training (6) — Programs that upskill government staff or practitioners already in the field.

Government

• Government (38) — Federal, state, and city digital services offices, innovation labs, and technology transformation teams operating inside government.

Capital

• Philanthropy (32) — Foundations, funder collaboratives, and pooled vehicles whose primary activity is grantmaking in support of government effectiveness and public interest technology.

• Investor (14) — VCs, PE firms, family offices, and impact funds with an explicit and demonstrated focus on government technology that builds capacity.

Industry

• GovTech (18) — Mission-oriented technology companies whose primary customer is government and whose product increases operating capacity rather than substituting for it.

• Digital Services & Consulting (15) — Mission-locked organizations whose output lands inside government: human-centered delivery, technical assistance, and process redesign.

What the map is for

This is a multi-modal information marketplace that acts as State Capacity’s connective tissue. A few examples.

You work inside government and you’ve identified a real problem. You need to figure out which startups sell into your tier of government, which digital services firms can help you implement, whether a fellowship might embed someone on your team for a year to help, and who has already tried a version of this.

You’re a founder building for government. You need the investors who actually understand government sales cycles, the communities where your future champions hang out, the state and city teams most likely to pilot something new, and the fellowships that might become your talent pipeline.

You’re a technologist looking for your next role in this space. You need to identify digital services firms that are hiring, state and city innovation teams that are growing, fellowships for experienced operators, and advocacy organizations that need your expertise.

All in one place for the first time.

The call to action

Browse the full database: Review it. Filter it. Find your organization. Find one to partner with.

Improve the map: If an organization should be on the map and isn’t, request to add it. If I made a mistake or you have further thoughts, post a comment.

Where this could go

This is version one.

The next layers are individual contacts, live problem statements, and profiles of people looking for their next role.

Perhaps in tandem with events to unlock information flow: VC + Advocacy sponsored AI-hackathon nights, with a Training partner facilitating and a Fellowship organization on the hunt for new talent.

A full-fledged community-driven resource focused on unlocking State Capacity delivery. But it only happens if the community helps deliver.

Let me know your thoughts.

This is part admonishment of how we think about AI adoption and part guide to selling AI better.

AI adoption isn’t technical, it’s human

The prevailing sentiment in Tech is that adoption is a technical problem: map the process, redesign it, automate it, deploy. Maybe that works at the bleeding edge. At most companies, that leads to a perfect solution with no uptake.

Why?

If you’re deploying AI, you’re inevitably going to run into mission-critical people who simply do not use, or even resist, AI. Until you win them over, your project is dead.

Tech focuses on rigorous discovery, thorough testing, and complex training. But non-users aren’t convinced by process. Or it indexes towards a “sink-or-swim” model. But that’s not great for morale anywhere, and I’m not convinced it actually leads to meaningful productivity gains - certainly not in government.

What does work? Getting non-users to “aha” moments that demystify technology. But until people trust a tool, they won’t use it. Technical capability is irrelevant.

If you’re selling AI, it’s your responsibility to get people over that hump; to teach them to trust not just your tool, but AI in general. Trust is AI’s currency. It doesn’t sell without it. You have a fiduciary duty to build trust.

Trust is all about communication

Whether CxO or an analyst, building trust in AI requires answering four questions.

1. How does AI work?

2. Is it good?

3. Should I be worried?

4. What do I do?

AI users and non-users don’t speak the same language, and answers are fraught with mistakes. So, I’ve demystified the conversation for you.

Mistake #1: You answer the question directly instead of understanding what folks are really asking.

Q1: How does AI work?

Insight: Folks are not asking about AI’s fundamental building blocks, they want to know if they can trust it. Frame AI in a situation that is compelling and relatable.

My answer: Think about AI as if it were a new employee just out of college. They’re probably pretty smart. They have baseline capabilities but lack expertise. You ask them to build a model. How do they do that? You give them access to Excel. Maybe you share model templates and firm best practices. You give them detailed instructions covering exactly what you need. And you let them iterate and learn, getting better over time.

AI is the same. Your AI model has (1) some amount of reasoning and language processing capabilities, which we call intelligence. And (2) it can access tools and follow instructions, which we call context. It can even iterate and learn. The more you learn to use the tool - better understanding of underlying capabilities, more specific phrasing, wider context offered - the better the output.

Mistake #2: You use technical jargon when simplicity is more effective.

Q2: Is it good?

Insight: They’re pushing for a line in the sand to define AI capabilities in the context of their work. You must balance theory with pragmatism and craft an answer that feels real.

My answer: AI has a bunch of basic capabilities that might apply to any process. It might be good at research, or information synthesis, or data analysis, or coding. How “good” it is depends on how well those capabilities map to the problem you’re solving. Now if you want to define whether AI intelligence is closer to analyst work or VP work, you’re going to have to play with those capabilities yourself. But what I think is [insert whatever you think - after all, you do have to answer their question directly!].

Mistake #3: You lack empathy.

Q3: Should I be worried?

Insight: This is a very human moment; a mixture of fear of the unknown and fear of falling behind. Approach with empathy, not theory.

My answer: I don’t know. Realistically, in terms of intelligence, benchmarks show that LLMs are already pretty darn good. Then it’s about context. People are building massive libraries of prompts to describe work, scripts to give AI standardized tools, and even armies of pre-built AI tools (you might have heard them called agents) that can complete large tasks. Over time, chances are we’ll have smart enough models, enough prewritten context, and enough prebuilt agents to complete a lot of work as we know it today. But, does that mean humans will be replaced? Maybe? We still need people reviewing AI output. And just because work is more efficient doesn’t mean there’s less of it - in fact, historically it has meant the opposite. I guess we’ll see.

Mistake #4: You misread the moment and sell too soon.

Q4: What do I do?

Insight: Note that the question is “I”, not we. AI continues to feel deeply personal even in a professional context. This is an opportunity to pitch, but not a detailed deployment roadmap. It is just conversation 1.

My answer: I feel like the key is just getting familiar with AI. I know you’re quite busy, but could we find some time to sit and play with a few tools? Are there use cases you think are really painful and repetitive and AI might be valuable? Or maybe we can work with folks who are excited about AI and problem solve with them?

The Final Word

The cultural zeitgeist is AI replacement. My practical experience is that humans make AI effective. If you want to sell AI, start by building trust with humans.

Because we’re in the age of AI, instead of an article, I’m writing a prompt.

Who You Are: An expert at large enterprise AI deployment; familiar with all the complexities of people, process, and systems that slow deployment, but with unrelenting optimism that AI offers a chance to skip past decades of sluggish technical adoption and upkeep.

Your Mission: Help me address real state capacity issues by suggesting an internal AI deployment strategy for a hypothetical government agency.

• In Scope:

  • Internal productivity improvement

• Out Of Scope:

  • Policy generation

  • Constituent services (chatbots can be done well and not so well – but not my focus)

  • Non-tech considerations (procurement and hiring can be tough – but let’s paint the skyline before designing the bricks)

Capabilities: We have simplified AI capability to three non-exclusive stages of adoption maturity; feel free to tell me there are better approaches or a more expansive set of options.

• Basic Large Language Model: Source a standard Enterprise LLM with suite of GenAI capabilities and offer basic training on how to use it **Hypothetical org may or may not be here

• Specialized AI Tools: Deploy new tools that augment specific business processes without significant system customization (e.g., Cursor for coding, Glean for file search)

• Custom AI Tools: Explore opportunities for customized, multi-step automation and AI-driven decision support across systems and workflows.

Considerations: As with most large organizations, our hypothetical org has meaningful barriers to AI adoption; some may be unique to government.

• Lots of data stored across many systems, both on-prem and cloud

• Complex file storage system spanning multiple storage locales

• Incentivizing employees to learn is appropriate; demanding uptake without support is not

• Regulatory and privacy concerns are of the utmost importance

Questions: We think these are the key questions, but feel free to tell me differently – or push for more information to inform answers.

• How have other organizations approached AI deployment?

  • What is the right way to think about AI capabilities?

  • What are common pitfalls and lessons learned in AI rollout?

• How does one prioritize top AI use cases?

  • Cross-organizational vs. function-specific use cases

  • Prioritization scorecard e.g., attractiveness (workload intensity, repeatable process, mission importance) vs. ease of implementation (staff interest, uniqueness of work)

• What are common models for tackling AI deployment?

  • Standing up deployment team(s) for intensive process-mapping and AI solutioning

  • Self-service enablement through curated playbooks, templates, and guidance materials

  • Train-the-trainer or internal champion programs

  • External consultants support intensive deployments

• What resourcing (staffing, skill sets, budget, timelines) are appropriate to support AI deployment?

  • How does this differ by deployment model?

• What other work must the org do in parallel to enable AI?

And the most important question

How do we make AI deployment scalable? Is intensive workflow and process mapping charted to AI capabilities the only way to maximize productivity (and is that really the existential AI question)?

Closing thoughts

• My commitment to you: Share a follow-on article summarizing my learnings.

• Please note: None of my words or views represent any official government or agency position.