Presented by Granicus & Carahsoft
Artificial intelligence is creating new possibilities across state government, but New Jersey is taking a deliberately practical approach to the technology. In a conversation with State Gov Today’s George Jackson in Trenton, Dave Cole, Chief Innovation Officer for the State of New Jersey, explains why successful AI adoption begins with understanding the problem—not selecting the technology. From accelerating business permits to helping state employees work more effectively, Cole says innovation must ultimately produce measurable improvements for the people government serves.
For Cole, the most important question about artificial intelligence is not what the technology can do. It is what problem New Jersey needs to solve.
That distinction guides how the state approaches AI, digital services and public-sector innovation. Cole’s team begins by studying the experiences of residents, businesses and state employees. Only then does it consider which technology—or combination of technologies—best fits the challenge.
“Understanding the problem for us is really the most important part,” Cole said. “Then we can start to think about how do we solve that.”
AI has become an increasingly useful part of the state’s toolkit. Cole sees opportunities to analyze information, improve documents, accelerate routine reviews and build prototypes that help agencies test ideas. But he does not treat AI as the automatic answer to every government challenge.
Instead, New Jersey’s model follows a broader discipline: understand the user, build an initial solution quickly, put it in front of real people, gather feedback and continually improve the service.
Making permitting more transparent and efficient
One example is New Jersey’s permitting dashboard, an application designed to make it easier to do business in the state.
Businesses often need permits from several agencies, making it difficult for applicants to understand where their requests stand or what may be delaying them. Individual agencies may also lack visibility into dependencies involving other parts of state government.
The permitting dashboard brings that information together. Applicants can follow the progress of their permits, agencies can see where applications sit within their own processes and state leaders can identify cross-agency dependencies that may be causing delays.
New Jersey launched the dashboard in June with 10 pilot permit projects spanning three agencies, including those responsible for environmental regulation, transportation and community affairs. The initial goal is to improve transparency and accountability, but the dashboard also gives the state a foundation for improving the permitting process itself.
Permit reviews frequently require employees to examine large document packages, confirm that the required materials are present and extract information for further review. Cole said AI can help scan documents for completeness and pull key details into a spreadsheet or another format that is easier for employees to evaluate.
That does not eliminate the need for an expert review. It reduces the amount of time employees spend on repetitive document handling so they can concentrate on questions requiring experience and judgment.
The permitting initiative complements business.nj.gov, the state’s one-stop resource for people starting or expanding a business. The website provides step-by-step starter kits for ventures such as restaurants, food trucks and beauty salons, helping users understand the registrations, licenses and permits they need.
According to Cole, businesses that use business.nj.gov begin operating an average of four weeks sooner than businesses navigating the traditional process. More than 60,000 businesses have used the resource since its launch.
The two efforts address different parts of the same experience. Business.nj.gov helps people determine what they need to do, while the permitting dashboard provides greater visibility after an application enters government. Together, they demonstrate how digital service design, process reform and AI can contribute to meaningful economic outcomes.
Giving state employees a safe way to learn AI
Cole also sees adoption by state employees as an essential part of New Jersey’s AI strategy.
AI tools are already being incorporated into many commercial technology products, and employees will naturally begin experimenting with them. If government does not provide training, clear rules and an approved environment, employees may turn to consumer tools that were not designed around public-sector security, privacy and accountability requirements.
New Jersey responded in 2023 by developing AI awareness and readiness training for state employees. The approximately 90-minute program is designed specifically for public professionals and uses examples that reflect the situations government employees may encounter in their daily work.
The training establishes a baseline around privacy and security while helping employees understand the limitations of AI. Those limitations include hallucinations, in which an AI system produces inaccurate information, and the potential for bias based on the data used to train a model.
New Jersey paired the training with the New Jersey AI Assistant, providing employees with an approved place to apply what they learn.
When a training lesson asks participants to draft a memo from a set of bullet points, for example, employees can try the exercise directly in the assistant. They do not have to imagine the result or create a private account on an outside platform. The assistant can also provide reminders about responsible-use guidelines as employees work.
The training and the assistant are designed to reinforce each other. Employees gain hands-on experience, while the state retains the ability to update its guidance as the technology evolves. The objective is to build a workforce that can incorporate AI into its daily work while understanding the responsibilities that accompany its use in government.
The training has also reached public servants outside New Jersey through InnovateUS, a nonprofit learning consortium launched by Cole’s predecessor, Beth Noveck. Broader adoption expands access to public-sector AI education while providing feedback about the questions and concerns government employees have as they begin working with the technology.
Turning a successful pilot into a statewide platform
New Jersey’s approach to scaling innovation begins with what Cole describes as “show, don’t tell” or “demos, not memos.”
His team uses research and lightweight prototypes to transform an idea into something agencies and users can see, test and evaluate. People often find it easier to react to a working concept than to envision a service based only on a memo or presentation.
AI can help teams reach that stage faster. It may accelerate the creation of a prototype even when the final production system will later need to be rebuilt to meet government requirements for security, reliability and scale.
A document-validation project with the New Jersey Economic Development Authority illustrates how that process can move from a focused pilot to a reusable platform.
Applications for small-business grants, loans and other programs require applicants to submit supporting documents. If someone accidentally uploads an outdated or incorrect file, that mistake may not be identified until an employee reviews the application.
During periods of high demand, Cole said it could take two weeks before an agent reaches the package. If the applicant then submits another incorrect document, the process could be delayed again. Several exchanges can add a month to the time required to complete an application.
The state prototyped a tool that can flag a potentially missing, outdated or incorrect document. In its most basic form, it gives an agent immediate feedback before a full review. The same capability could also alert applicants when they upload a file, allowing them to check the document before submitting the application.
Once the concept demonstrated value, Cole’s team recognized that the same problem existed across government. Many programs collect forms, manage application queues and lose valuable time when the information they receive is incomplete.
The team began separating the Economic Development Authority-specific components from the core capability and turning the pilot into a platform that other agencies could use.
The emerging system uses generative AI to examine sample documents for a program and help develop validation rules. Those rules can then be saved as deterministic checks that can be applied consistently to future submissions.
That combination is important. AI helps an agency configure a new program more quickly, while conventional code performs the repeatable validation checks. The platform could eventually support dozens of programs without requiring each agency to develop an entirely new system.
Measuring innovation by its effect on people
Cole defines innovation as finding new ways to solve difficult problems. But experimentation alone is not the objective.
The work must deliver a result for a defined group of people, and the people who will use a service should be involved throughout its development. That emphasis changes how the state measures success.
A polished prototype has little value if eligible residents never discover it or cannot use it. New Jersey’s innovation team includes professionals with private-sector marketing experience who apply those skills to public programs. They help residents learn that a benefit exists, determine whether they qualify and complete the steps needed to receive it.
Adoption is therefore not simply a communications metric. It is part of delivering the service.
For the team, the ultimate result may be a resident obtaining health insurance, receiving food assistance or discovering an opportunity to afford a home. Those outcomes also give public servants a direct connection between their work and its effect on people’s lives.
AI can expand what is possible, particularly when government needs to analyze large data sets or match records that do not align perfectly. Machine learning can identify likely connections and assign confidence scores much faster than a manual process.
But the technology remains part of a broader strategy that includes policy, research, digital service, communications and program operations.
Looking beyond the excitement surrounding AI
Cole acknowledges that the excitement surrounding AI can be useful. It can challenge the longstanding assumption that certain government problems are too difficult to change and encourage employees to think differently about backlogs, complicated processes and service delivery.
That same excitement can also create pressure to purchase a product before an agency fully understands the problem it needs to solve.
New Jersey must examine where AI is actually being used, whether a vendor can support the capability it promises and whether the product could obscure decision-making or introduce bias and inaccurate information.
Sometimes the best answer is not an AI chatbot.
If residents cannot find information on a government website, the first questions should be whether the content is current, written for the intended audience and organized logically. Better navigation, search and writing may solve much of the problem.
AI may still contribute by simplifying language, improving readability or producing a first draft of a translation that is then reviewed by a person. But the technology supports the service rather than defining it.
For Cole, the critical question is where AI belongs within the overall process. Improving the underlying content and information architecture also gives any future chatbot or virtual assistant more accurate information to work with.
Keeping people accountable for consequential decisions
As AI becomes more deeply embedded in government workflows, New Jersey is drawing a firm line around consequential decisions.
Cole called human judgment “a non-negotiable.”
When a decision affects someone’s access to a benefit, awareness of a program or another important outcome, a public employee remains responsible for that decision. AI can simplify and organize the information used to reach a conclusion, but it does not remove the government’s accountability for the result.
The document-validation project reflects that principle. The system is not intended to automatically reject an application because it believes a document is incorrect.
Instead, it can tell an applicant that a document may be outdated or incomplete and invite the person to review it. The applicant retains the ability to continue. Likewise, a state employee can use the tool to organize information while still applying professional judgment and approving the final determination.
That balance captures New Jersey’s broader approach to AI. The state is moving quickly enough to experiment, learn and build, but deliberately enough to protect accountability.
The process begins with real public problems, involves the people affected by them and scales solutions only after they demonstrate value. AI may make entirely new forms of service possible, but New Jersey’s measure of success remains distinctly human: whether government produces a better outcome for the people it serves.
About The Future Public Servant
This interview is part of The Future Public Servant, a State Gov Today series presented by Granicus in partnership with Carahsoft. The program explores how public-sector leaders are preparing their organizations and workforces for a future shaped by artificial intelligence, changing citizen expectations and new models of digital service delivery. Individual interviews will be featured on TheFuturePublicServant.com and brought together in special episodes of State Gov Today.