Oakland’s AI Experiment: Testing Before Buying It

What happens when a city invites innovators to help reimagine how government works with artificial intelligence? Oakland is about to find out.

The city issued a request for information (RFI) recently, calling on AI companies and academic researchers to propose real-world solutions to some of its toughest operational and community challenges. The initiative — coordinated by Oakland’s AI Working Group — represents a new phase in how the city approaches technology.

“The RFI has 30 curated use cases,” Tony Batalla, the city’s CIO, said. “We’ve asked companies and researchers to respond to those and identify ones where they feel it’s a strong fit with what they’ve developed — and then work on that specific use case.”

Each of the 30 use cases originated from city departments themselves and spans nearly every aspect of city life — from AI-based accessibility tools that fix digital documents for Americans with Disabilities Act compliance, to virtual assistants that help residents navigate permits and public services, and AI-powered analysis of body-camera footage to improve transparency. Other ideas include real-time multilingual translation, predictive analytics for staff scheduling, AI monitoring for illegal dumping and digital twins for city planning.

Batalla explained that staff from “20 or so departments” submitted their own challenges and ideas to form these categories.

“There’s a lead staff person behind each one of those,” he said. “If a proposal is submitted for a use case, it will be evaluated by that staff person. If they feel like they want to move forward with a pilot engagement, then we would go ahead and execute that.”

The pilots will then run 16 weeks, structured around a newly created memorandum of understanding (MOU) template developed with the City Attorney’s Office. It’s a document Batalla described as critical to both efficiency and accountability.

“We created the MOU for the pilot engagements, which still includes protections for the city — data security terms, success measures, statements you’re trying to solve and how you’ll measure success,” he said. “That’s built into the MOU. Over that period, we can determine whether or not it’s successful, and the successful ones we would look to fully fund and scale citywide in our future budget.”

For Oakland, the pilot program essentially offers a controlled way to explore new technologies without committing taxpayer dollars up front. Batalla called it a deliberate, low-risk approach to innovation.

“We’re very interested in demonstrating this as a safe way to engage on some of these cutting-edge technologies where we can prove that they work before we buy something,” he said. “We can prove that they’re safe or that there’s some security built in before we make a purchase.”

The AI Working Group, co-led by Batalla and Deputy City Administrator Joe DeVries, includes representatives from nearly every city department. The team was created in 2024 through the city administrator’s office to ensure citywide participation and executive support because, as Batalla put it, they wanted to “have administrator-level representation and backing with AI being a strategic priority for the city.”

Over the past year, the group has built the city’s AI foundation, including an equity framework, security guidelines and a data governance pilot that prevents sensitive data from being accessed by AI tools. Building on that, roughly 50 Oakland employees are now piloting Microsoft 365 Copilot, exploring how the platform can support their daily work and departmental initiatives. According to Batalla, these employees not only test the tool’s governance features, but developed the ideas behind their projects and secured approval to implement them.

In partnership with Northeastern University’s Oakland campus, the city is also studying how employees interact with AI in real time. A graduate student is doing a research study around the city’s rollout of data governance and Copilot, Batalla said, and will take on additional use cases next semester — possibly building a solution.

For Batalla, all these efforts tie back to the city’s three-year strategic plan and its broader push toward digital transformation. Many of the AI use cases, particularly in areas like contracting and procurement, he said, address challenges common to cities everywhere. By exploring how AI can simplify these processes, the CIO said, Oakland is taking the work it’s doing in AI and directly aligning it with the strategic priorities it has already identified.

While Oakland is open to working with major tech companies, Batalla emphasized that officials hope to hear from smaller firms and academic innovators as well during the RFI process.

“We do invite large companies to respond, but we’re also really interested in startups and researchers,” he said. “They may have something unique — a bigger company is going to bring its whole platform, but smaller companies can sometimes be more agile.”

Ultimately, Oakland’s approach to AI is not focused on chasing the latest tech tool, but on building trust and understanding.

“We’re kind of in a phase where we want to unleash some of the potential and really sort out what’s real from what’s hype,” Batalla said. The RFI process, he said, reflects this approach and provides a structured way to safely test new technologies and determine whether they truly deliver on their promises.

This story was originally published by Government Technology, Industry Insider — California's sister publication.