AI-powered permitting is speeding up solar deployments in California

At the start of September of this year, John Caprarelli, a city building official for Santa Clarita, California, had a deadline to meet. By the end of the month, his city of about 220,000 people had to start issuing ​“instant permits” for all residential solar and solar-battery projects — as did every city and county in California with more than 50,000 people under state law.

Santa Clarita had planned to use SolarApp+, a solar permitting software platform developed by the Department of Energy, to hit that target. But on the advice of a colleague, Caprarelli called up Leila Banijamali, CEO of San Francisco–based startup Symbium, to look into her software as an alternative.

“I was a little sheepish to call Leila — I was calling her on September 1^st to have it up and running on the 30^th,” he said. ​“Most agencies would take 12 months to implement a permitting system.”

But by the end of the month, Symbium — a spinout founded by Banijamali and two scientists at Stanford University — had its software up and running for Santa Clarita. That made it the 22^nd city in the state to take up Symbium’s offer of a free pilot implementation to meet California’s instant-solar-permitting deadline.

“To get a permit system launched within a month of contacting them — it’s unheard of,” Caprarelli said.

Symbium is part of a wave of software companies looking to speed up notoriously sluggish permitting timelines for renewable energy projects. These startups say approval processes can be tightened up dramatically with the help of automation software.

California’s state law mandating instant permitting, passed last year, more or less forces its major cities to adopt this technology, which could eventually help officials around the country process clean-energy permits at a pace that’s more in line with the urgency of the climate crisis.

Before it adopted Symbium’s platform, a Santa Clarita permitting employee would spend up to an hour per application to check each of the hundred or more solar applications that the city receives per month to ensure that the contractor is licensed, that the project meets electrical and fire codes, and that every piece of equipment involved is certified by the state.

Today, ​“the solar contractor can enter their information about the solar system they want to build, and Symbium will ask the questions and gather the information and verify if the system meets code requirements,” he said. If a contractor’s plan is out of compliance, Symbium alerts them with a pop-up window that explains the error and how to correct it.

Santa Clarita is able to pass the staff-time savings from this automation back to contractors in the form of a roughly $120 discount on the typical $450 permit processing fee, Caprarelli said. Symbium collects $25 per application from the contractor, but that still leaves about $100 in savings per permit application.

Not only is Symbium fast, but it’s also flexible. ​“A lot of the projects we receive are additions to existing solar systems,” such as adding more solar panels, inverters or batteries, which Symbium can handle, he said. It can also integrate local codes that differ from the norm, such as the Los Angeles County Fire Department’s special requirements for lithium-ion battery installations, he said.

So far, the on-site inspections of the projects instantly permitted via Symbium have checked out, he added. ​“If it was producing designs that didn’t meet code — equipment that wasn’t properly listed, wire sizing that wasn’t in compliance — our inspector would see that,” he said.

By replacing the need for a human to do these simpler permit-inspection tasks, the software ​“allows our staff to focus on the more complex issues that require that human attention,” he said. ​“Now we can put our staff time into things that add more value.”

An AI that can interpret the law

The idea of automating the grunt work of interpreting legal codes wasn’t new to Banijamali when she co-founded Symbium in 2019. Back in 2014, she launched another company, Startup Documents, that automated some of the more onerous and repetitive work she did as a lawyer representing startups.

That work ​“caught the attention of folks at Stanford — particularly at CodeX,” a joint program of Stanford Law School and its computer science department, where she won a fellowship in 2017. That’s where she met her co-founders, Symbium CTO Abhijeet Mohapatra and Chief Scientific Advisor Michael Genesereth, who were in the midst of developing what Banijamali described as ​“a major technological breakthrough” in the realm of artificial intelligence known as ​“logic programming,” and its use in the field of ​“computational law.”

Computational law focuses on building ​“computer systems capable of assessing, facilitating, or enforcing compliance with rules and regulations,” Genesereth, a Stanford computer science professor, wrote in a 2021 paper — and, importantly, that can ​“apply regulations to real or hypothetical cases without additional input from human legal experts.”

Banijamali cited one well-known version of a similar type of computerized legal analysis: TurboTax, the tax-preparation platform. TurboTax calculates a taxpayer’s inputs ​“against the laws and requirements to tell you, ​‘Here are your tax obligations for this year,’ so you don’t have to read the IRS code manually,” she said.

But TurboTax has been developed through years of work by tax law experts and software engineers, and it must be updated by the same costly manual work with every change of the tax code, she said. Symbium’s computational-law technology, by contrast, applies the techniques of logic programming to convert legal codes and regulations into ​“logical language” that computers can interpret, Banijamali explained.

“It takes a little longer to implement — but the benefit is that once the first instance is complete, it’s very easy to scale it up to other systems.”

Mohapatra, who earned a doctorate in computer science at Stanford, explained that logic programming is different from other branches of AI, such as large language models like those popularized by ChatGPT, because those ​“are probabilistic systems.” While they can create facsimiles of writing or artwork created by humans, their outputs are ​“essentially guesswork at this point,” which isn’t good enough for legal work.

Computational law, by contrast, can be used to codify what he called ​“non-discretionary law” — the parts of legal codes and regulations that aren’t subject to multiple alternative interpretations, or ​“the portion of the law that is zeroes and ones,” and deliver ​“100 percent accurate” results.

Creating a logic program does require humans to translate laws into logical language the computer can understand, Mohapatra said. But ​“the good news is that logic programming, unlike traditional programming, is very easy” to learn for people who aren’t software engineers but are domain experts in their fields — solar engineers or code-compliance experts, to name two groups of experts Symbium has enlisted to work on building its core body of knowledge.

Plus, those humans don’t have to keep feeding Symbium’s computational logic system forever, he said. ​“How do we encode regulations and maintain it at scale? To encode regulations as logical sentences, we start by finding common patterns or structure in these regulations,” he explained. ​“Think of these patterns as templates or blueprints, also known as schema. This makes it easier to encode regulations that are tailored to specific jurisdictions, as well as updates to these regulations, by fine-tuning the parameters within the schema.”

In other words, once a master template is in place, tweaking it to accommodate local code discrepancies or newly mandated changes to regulations becomes much simpler. That’s the secret to building an automation platform that can both deliver accurate interpretations of law and scale up to serve an ever-increasing number of tasks without overburdening the humans involved, he said.

Meeting the need for instant solar permits 

That flexibility is what allowed Symbium to quickly pivot from its initial applications of streamlining building-permitting processes for accessory dwelling units, or ADUs — the small homes built in people’s backyards — to speeding up solar projects. The spur for that was SB 379, the 2022 law that mandated instant permitting across California, starting this year for the most populous cities and counties and extending to all others in late 2024.

Symbium quickly launched a solar-permitting pilot to put its software to work on that task. In late 2022, it landed the city of Vacaville in California’s Central Valley as its first instant-permitting partner.

When SB 379 was passed, ​“A lot of us were asking, how will this work?” said Aleris Dunn, Vacaville’s chief building official. The California Energy Commission’s guidance on how to comply with the law only listed SolarApp+ as a software platform available to fix those kinds of problems. But, she wondered, ​“How will it determine code compliance? How will it achieve real-time permit issuance and plan-check?”

Vacaville was already an outlier among California cities and counties in offering instant permits for solar projects, she noted. But that process didn’t include a plan review by a city employee prior to a building inspector being sent out to review the installed project. That meant that ​“if something was installed incorrectly or undersized or oversized, the contractor had to redo the installation.”

Dunn and her colleagues met with the DOE team that helps cities and counties implement SolarApp+. ​“Could it be tooled to meet the demands of the jurisdiction? Yes, it could,” she said. But then she met with Symbium and saw that the startup’s product offered a deeper analysis of a permit application’s compliance than what SolarApp+ was providing.

That’s not the fault of SolarApp+, which is currently used by more than 100 cities and counties to streamline residential solar permits, Mohapatra said. ​“Make no mistake — it’s a great tool” for automating much of the permit application process. ​“But that alone is not enough.”

And without the AI structure that Symbium uses, it’s difficult if not impossible for a small team of software engineers to continually update and upgrade a platform to keep up with the ever-evolving complexities of the real world, he said.

That view was echoed by Barry Cinnamon, a longtime Silicon Valley solar entrepreneur and CEO of solar and battery installation firm Cinnamon Energy Systems. ​“SolarApp+ is getting pretty good universal traction for straightforward solar installations — rooftop projects, nothing really funky,” he said. ​“And they’re starting to get more traction with solar and battery combinations. But the batteries are so darn complicated with location and sizing and spacing…and you have to keep up with the details of all the equipment, which is constantly changing.”

But Cinnamon also expressed some skepticism that Symbium’s platform would be able to handle all of this complexity and diversity. ​“All of these master permit simplification processes — which are incredibly useful when they work — are going to need to resolve the devil in the details” of every different jurisdiction doing things a slightly different way, he said. ​“I suspect that the challenges that SolarApp+ has experienced in the solar business around the country will be reflected at Symbium.”

Mohapatra agreed that permitting agencies will need to prove that Symbium’s platform does what it promises in order for the company to gain further traction. ​“They need proofs of concept. Will it even work? The solar permitting pilot was the start of that relationship.”

Banijamali said that the quick-learning nature of Symbium’s computational law approach can handle the foibles and variations it’s encountering in its deployments thus far. In fact, ​“Symbium can handle all residential solar projects — and it’s very easy for us to add commercial,” she said. ​“It doesn’t matter if you’re adding to it, modifying it or building new. And Symbium handles hundreds of different permitting processes in the same way.”

This flexibility also opens up applications well beyond solar, Banijamali said. Beyond its solar-permitting pilot, the startup has also launched a broader ​“express permits” pilot with California cities and counties to automate a host of simpler electrification projects and installations of home components and appliances. And in July, Symbium joined the cleantech program of Third Derivative, a startup accelerator founded by think tank RMI and nonprofit New Energy Nexus, to explore ways to apply its technology to streamlining a wider array of electrification and decarbonization projects. (Canary Media is an independent affiliate of RMI.)

Permitting roadblocks are also a problem for installing home EV chargers and electric heat pumps, two other technologies that must be adopted at a breakneck pace to forestall the worst harms of climate change, she noted.

Symbium raised a $4.1 million seed round in 2019 and followed it up with another funding round earlier this year. ​“We haven’t disclosed this last round,” Banijamali said, but she did say the company raised more money than it had expected to. The startup plans to use the new funds to expand outside California next year, as well as to put its AI to work on this expanding array of permitting automation tasks, she said.

Symbium’s overarching goal is to create a ​“citizens dashboard” that streamlines ​“citizen-to-government interactions” in multiple ways, according to Mohapatra. But ​“we want to start with residential home-improvement projects,” he said. ​“Right now, the biggest need of the hour is decarbonization.”

With the passage of the Inflation Reduction Act last year, tens of billions of dollars in federal incentives are now flowing to rooftop solar, battery, EV charging and home electrification and efficiency projects, he noted. But if people can’t build these projects quickly enough, ​“these incentives will not be of much use.”