New Jersey’s AI Solar-Permitting RFI: Can Code Review Be Automated Safely?
New Jersey is not buying an AI permitting system yet. It is asking the market what a government-side platform would have to do before a state could trust it.
Focus
AI solar permitting
Written for
Solar installers, code officials, permitting software teams, and clean-energy operators
Editorial standard
Facts, status, and practical implications
At a glance
- New Jersey’s Department of Community Affairs posted an exploratory RFI on May 27, 2026.
- The notice is market research, not a procurement or a contract award.
- The useful question is how AI can reduce routine review while keeping code officials accountable.
The news is a question, not a contract
The New Jersey Department of Community Affairs has put a very current question in front of the solar industry: what would a government-side, AI-driven solar-permitting platform actually need to look like? The agency’s May 27 notice asks for information about the state of the art in automated permitting. That is a meaningful signal because it moves the conversation away from a homeowner asking whether a chatbot can fill in a form and toward a public authority asking whether software can support a regulated decision.
The wording matters. New Jersey calls the notice an exploratory Request for Information, and it says clearly that it is not a competitive procurement. No vendor has won a contract because the notice exists. No state-wide system has been approved. A responsible article should resist the temptation to turn a discovery exercise into a product launch story. The real news is that government is collecting evidence before it defines the requirements.
Why permitting is an operations problem
Solar permitting looks administrative from the outside, but the work contains many small decisions. A reviewer checks equipment information, electrical details, structural assumptions, site drawings, and local requirements. An installer may submit the same kind of project repeatedly, yet each jurisdiction can ask for information in a slightly different format. Missing a document, using an outdated rule, or sending a drawing that is hard to read creates another round of email and another place for a project to wait.
That is why automation is attractive. A system could sort an application, identify missing fields, compare a standard design against known rules, and route unusual cases to a person. It could also give applicants a clearer explanation of what needs to change. The benefit is not simply speed. It is less ambiguity for the installer and a more consistent starting point for the reviewer.
What a credible platform would need
A serious platform would need structured data, a maintained rules library, document handling, identity and access controls, and an audit trail. It would need to show which drawing, equipment record, or rule produced a flag. If a reviewer asks why an application was returned, the answer cannot be “the model was uncertain.” The answer should point to a missing value, a failed check, or a rule that the reviewer can inspect.
The system would also need to handle imperfect inputs. Plans arrive as PDFs, scans, photographs, exports from design tools, and files produced by different installers. Optical recognition can help read a document, but recognition is not proof that the document is correct. The platform should preserve the source file, identify low-confidence extraction, and make the reviewer’s correction visible. Accessibility, language support, retention policies, and secure integration with existing municipal systems belong in the requirements too.
Where humans stay in charge
Automated permitting should be treated as a controlled decision-support workflow, not an excuse to remove professional judgment. A code official still needs a way to review exceptions, unusual structures, conflicting information, and changes in local policy. The software can prepare a recommendation; the authority should remain visible. That is especially important when the same system is used by many jurisdictions with different practices and risk tolerances.
The safest early boundary is narrow. Routine applications that meet clearly defined conditions may move through a faster path, while anything outside those conditions is held for review. Every automatic outcome should have a reason, a timestamp, and a route back to a person. If the platform cannot explain itself, correct itself, and record who accepted the result, it is not ready to become part of a public permitting process.
What installers and software teams can do now
Installers do not need to wait for a state contract to improve their side of the workflow. They can standardise equipment data, naming conventions, drawing exports, revision control, and pre-submission checks. A clean packet makes any future permitting platform more useful. Software vendors can prepare by supporting open exports, clear validation messages, role-based permissions, and an evidence view that shows the source behind every recommendation.
The most useful pilot is usually not a fully automatic approval. It is a preflight assistant that catches missing information before submission and produces a review packet for the official. That approach creates measurable evidence—fewer returns, clearer reasons, and less back-and-forth—without pretending that a model should replace the person responsible for the code decision.
What to watch next
The next important development will not necessarily be a flashy demonstration. It will be the requirements that emerge from the RFI: which data formats the state expects, how local authorities would connect, what security controls are mandatory, and how the state would evaluate accuracy. Those details will tell the market whether the project is a narrow permitting aid or an attempt to build a broader digital public-service platform.
For now, New Jersey’s move is best read as a signal. AI is entering the permitting conversation, but credibility will come from evidence, not adjectives. The companies that understand that distinction will be better positioned when the questions become a specification and, eventually, a real procurement.
Sources and status
This is news analysis, not legal or financial advice. Proposals, forecasts, draft standards, and company statements can change. Check the linked primary sources and the publication date before relying on a detail.
From news to useful action
Need to make an energy workflow easier to run?
Remova Tech helps teams review opportunities, connect reliable data, and introduce automation with clear ownership and human control.