In my experience, there are many small business owners that think they need a Silicon Valley budget to see real-world gains from artificial intelligence. Usually, they look at me with a healthy dose of Northwest skepticism when I tell them otherwise. But lately, I’ve been pointing them toward an unlikely laboratory of innovation: Seattle City Hall.
In late 2025, Mayor Bruce Harrell launched the City of Seattle 2025-2026 AI Plan, a 26-page framework that moved the city from “thinking about AI” to running 40+ active experiments. For a small business owner, this plan isn’t just government paperwork; it’s a fine example of how to test-drive AI without crashing the car.
The Permitting Revolution: CivCheck and ArchistarAI
If you’ve ever tried to open a physical storefront or renovate a workspace in Seattle, you know the “permit crawl.” The city’s Permitting Accountability and Customer Trust (PACT) team decided to treat this as a logic problem rather than a paperwork problem.
They partnered with CivCheck and ArchistarAI to create an AI-driven pre-screening process. Instead of waiting weeks for a human reviewer to tell you that you forgot a signature or miscalculated a setback, the AI scans your application instantly. It catches errors before you hit submit.
The Lesson: This is “AI as a Tutor.” For small businesses, the takeaway is clear: Use AI to audit your own internal processes, like checking invoices against contracts or verifying shipping manifests, before they reach a human’s desk. This step helps to ensure that the expert isn’t wasting time on typos.
Safety and Infrastructure: Beyond the Hype
The city is also experimenting with “Invisible AI.” The Seattle Department of Transportation (SDOT) partnered with C3.ai and Microsoft to analyze “near-miss” collision data across 7,800 intersections. By processing 250,000+ data points that would take a human analyst years to sift through, they reduced analysis time by 90%.
Meanwhile, Seattle Public Utilities is testing AI for “Pipe Assessment.” They use computer vision to scan miles of sewer and water lines for cracks. It’s the ultimate “dull, dirty, and dangerous” task that AI excels at.
The Lesson: High-value AI isn’t always a chatbot. Sometimes it’s just a very fast pair of eyes. Small businesses in logistics or maintenance can learn from this: AI’s best use case is often the massive data set you’re currently ignoring because it’s too boring to read.
The Human-in-the-Loop Reality Check
Not every experiment has been a smooth ride. Recently, the Seattle Police Department (SPD) faced scrutiny when it was discovered that some public communications were likely 100% AI-generated using tools like Amazon Bedrock, without the required Human-in-the-Loop (HITL) attribution.
The city’s policy is firm: AI can draft, but a human must sign off. Whether it’s the Zendesk chatbots handling resident inquiries or the Online Business Directory updates, the city mandates that AI-generated content be labeled if it’s substantive.
The Lesson: This is where my “candor over hype” rule comes in. If you use AI to write your customer newsletters or product descriptions, you owe it to your brand’s integrity to have a human sanity check the output. AI is a great co-pilot, but it’s a terrible captain.
The Small Business Blueprint
So, what does the “One Seattle Data Strategy” tell us?
- Don’t build, borrow: The city didn’t build its own LLM; it used enterprise tools like Amazon Bedrock and Zendesk.
- Focus on The Friction: The city targeted permitting because it was the #1 complaint. Where is the friction in your business?
- Data is the Foundation: You can’t have a Unified Resident Contact System if your phone numbers are in three different spreadsheets.
The City of Seattle is proving that AI isn’t about the future—it’s about making incremental improvements now. If the local government can figure out how to use a chatbot to help a contractor understand the Seattle Municipal Code, you can certainly use it to help your customers find the right product on your website.