Anthony Jancso noticed that San Franciscans—in the tech community and beyond—can't help but opine on the city's myriad challenges. So he decided to do something about it: organize a hackathon with the aim of using AI to solve the city's problems.
He’s part of the tech-focused nonprofit Accelerate SF, which is hosting a two-day hackathon the weekend of Nov. 4 with an unorthodox mission.
“The core mission here is to just build the technology that solves problems in San Francisco,” Jancso said.
The event, which has buy-in from local politicians like state Sen. Scott Wiener and Supervisor Joel Engardio, is part of a growing trend of San Francisco-based AI experts and companies turning their attention to the problems of municipal government.
The city itself is already implementing AI and machine learning in some departments, even if policies around the use of generative AI—technology that can create images, text and video—are still in the works.
The Assessor-Recorder's Office, for instance, uses a machine-learning model to predict property values and identify properties in need of appraisal. The 311 mobile app also harnesses AI technology to guide users to the correct service request.
But other cities are ahead of San Francisco when it comes to experimenting with artificial intelligence, and they are tapping companies just a hop-skip away from San Francisco City Hall like Hayden AI, Abnormal Security and Glass. Those startups are tackling tasks ranging from optimizing public transit to securing government data.
Making Buses Better With AI
Hayden AI was inspired by CEO and co-founder Chris Carson’s experience riding a Muni bus. He noticed the driver was forced to manually record whenever a car stopped illegally in a bus zone. There had to be a better way, he thought.
The San Francisco company has developed an AI-based camera technology for government-owned vehicles like buses to automatically catch scofflaws. In Philadelphia, a pilot involving seven bus lines highlighted 36,392 violations over a 70-day period, potentially totaling millions of dollars in fines. (The city did not ticket any offenders during the pilot.)
But there is a possibility that the technology could be adopted more broadly across bus lines for Philadelphia’s transit agency, called SEPTA. The city recently passed a bill permitting the use of automated cameras to enforce violations on bus-only lanes, using data from the pilot as evidence of the need.
“Although 19% of vehicles were [caught], repeat offenders constituted 77% of violations, so we have some power users, you might say,” said Matt Zapson, a planner for SEPTA. The study found more than 10,000 bus riders were negatively affected by the blocked lanes, which delay buses and distract drivers.
Jenna Fortunati, Hayden AI’s communications manager, said the company is not currently in conversations to upgrade San Francisco’s manual system—which it has had in place since 2014—but hopes to eventually work with the city.
Hayden AI has proposals to incorporate its computer vision-embedded cameras into other municipal vehicles like street sweepers, to automatically ticket vehicles blocking their routes.
Cameras installed on buses could be used to also detect parking violations in bike lanes, a long-standing problem in San Francisco, Fortunati added.
Sounding the Alarm
Local governments are often low-hanging fruit for hackers with sprawling systems and limited IT budgets. This year, for example, the City of Oakland suffered a ransomware attack that led to a trove of private information being released onto the dark web.
San Francisco-based Abnormal Security uses AI to keep tabs on employee computer use for anomalous behavior—signs that a hacker has penetrated the system. The technology compiles a baseline of a user’s behavior in email, Slack and other workplace apps—and sounds an alarm whenever it notices something is off. The organization can then block access or force login changes, among other security measures.
Tas Jalali, head of cybersecurity at East Bay transportation agency AC Transit, said using Abnormal’s technology, his team was rapidly able to block account takeover attacks and phishing efforts. Outside of that more active role, Jalali said the software can filter through spam and “graymail” that can gum up employee inboxes, saving at least 120 employee hours a month.
The challenge of introducing its product to local government agencies boils down to the company’s software playing nicely with existing requirements and systems. For instance, AC Transit uses Microsoft’s email software—a common corporate standby. But some government agencies use antiquated email software that is incompatible with Abnormal’s technology.
The price of doing nothing is growing. According to IBM, the average cost of a data breach in 2023 was $4.45 million, a 15% increase since 2020.
“There's a tremendous amount of potential damage that can be done when educational institutions or health care institutions that are affiliated with the city are hacked,” said Dan Shiebler, the head of machine learning at Abnormal Security. “Defending against that often means incorporating artificial intelligence.”
Bringing E-commerce to City Hall
Glass is a San Francisco startup that is trying to take the consumer e-commerce shopping experience and bring it to local governments.
Founder Paola Santana’s pitch? Local governments have significant purchasing power, much of which is non-contract spending. Glass, she said, is aiming to simplify that process by merging the convenience of Amazon or Google shopping with the necessary mandates that government purchases require.
AI is built into the product to help government purchasers fulfill their particular needs, which can vary vastly from city to city. Some cities may have local business or sustainability mandates, while others may simply want the lowest costs to stretch their budgets further. San Francisco, for instance, has a “green procurement” initiative, emphasizing the purchase of eco-friendly products in its government purchases.
“When we go and source that product in real-time from vendors, most times we find between 5% to 30% off prices than what the government would have paid,” Santana said. “It's streamlining the process and getting the best value for taxpayers’ money.”
In Seattle, where the emphasis is buying from women- and minority-owned businesses, Glass was able to create a one-stop shop for city employees with products from multiple sources, which it boasts has saved time and taxpayer money.
Closer to home, Santana says Glass is in talks to roll out its product with San Francisco’s Office of Workforce and Economic Development and the Department of Public Works, among other departments.
“We just think the San Francisco government being more exposed to this technology, seeing what is being built and having more conversations with tech startups will unlock many problems that they've been dealing with for many decades,” Santana said.
That’s part of the hope of next month’s Accelerate SF hackathon: Pinpointing what problems can be addressed with artificial intelligence—and getting technologists and lawmakers to break bread in a public forum.
Already, they count buzzy AI startups Scale AI and Chroma, along with the Mayor’s Office of Innovation, as sponsors. Judges for competition comprise leaders in the emerging industry like Notion co-founder Chris Průcha as well as longtime public officials like City Attorney David Chiu.
By the end of the hackathon, Jansco hopes to find AI-based solutions to make tools like SF OpenBook, a “super hard to use” database of the city’s finances, more accessible to the masses.
“I hope that, in addition to the wonderful tools that will be made, it's also starting a conversation and a dialogue between tech workers and the city,” Jancso said.