RAIN — Week of September 28, 2026

Ask Amy

What are the most practical first steps a rural nonprofit or *small business* can take to use AI effectively when time, funding, and technical capacity are limited?    Sasha, Colorado

Great question Sasha, and one I get often. So what can we actually do with AI when there's no time, no money, and no one on staff who speaks fluent tech?

My advice is to start with a technical problem you're having. Is there a piece of software you already know you need and haven’t had the bandwidth to implement?

Maybe your online ad campaign isn't working the way you want. Someone told you there's a platform that would fix it. You looked into it once, signed up for the free trial, saw the dashboard, and closed the tab. Are you going to learn it this quarter? Probably not. Maybe over the holidays. Maybe in the shoulder season when things slow down. Then the shoulder season comes and goes and the tab stays closed.

This is where AI can earn its keep. AI can get you up and running on the software you already decided you needed.

Open the software platform. Open ChatGPT or Claude next to it. Describe what you're looking at and what you're trying to do. Ask why a setting exists. Paste in the error message. Ask it to walk you through the first campaign one step at a time, then ask it the question you'd be embarrassed to ask a consultant. Or ask it what usually breaks first and how to avoid common mistakes.

You don't have to become a technical expert. You're already the expert in your business or your mission. You're already the expert in your business or your mission: your customers, your season, your town. AI can get you to proficient, and proficient is all you needed.

Sasha, pick the one piece of software you've been putting off and use AI to finally get it up and running. Then get back to the work only you can do.

Until next week,

Amy

Every column starts with a reader question. Yours can be next: ruralai.org/rain

Stories of the Week

Inside our months-long investigation into Kevin O'Leary's Utah data center debacle

Stratos, the 40,000-acre, nine-gigawatt AI campus Kevin O'Leary pitched for Box Elder County, Utah, was approved by a state development authority before the county government was told. The Verge's investigation traces the backlash that cost two county commissioners and state Senate President J. Stuart Adams their seats.

Nine gigawatts is more than double Utah's average statewide power use, and county consent turned out to be the only real chokepoint. Rural officials asked to approve a project quickly should read how close that consent came to being routed around.

Senate Hearing on 'Rogue AI: Securing the Homeland Against AI Agent Attacks'

A Senate Homeland Security subcommittee heard testimony September 30 on the OpenAI agent swarm, roughly 1,200 agents trading 70,000 messages, that compromised Hugging Face. Dragos testified that an AI model independently steered an attacker toward a Mexican water utility's control network and attempted 2,800 credential guesses.

Only about one in 10 operators monitor their operational networks, and July's US water system breaches traced to unchanged default passwords at small utilities. Missouri's hospital and banker associations filed statements; credentials and OT visibility are the fundamentals small-town systems can fix now.

Microsoft goes quiet after church groups ask for 1% of data center costs

A coalition of about 25 Indiana congregations asked Microsoft to put 1 to 2 percent of the annual cost of its $1 billion Granger data center into a community fund, roughly $30 to $40 million a year. Microsoft approved up to $1 million in one-time nonprofit grants and went silent on the proposal.

Indiana's sales tax exemption on data center equipment can run 50 years and may cover $13.2 billion in purchases. We Make Indiana now wants county officials to withhold a proposed TIF district until Microsoft signs a binding fair-share agreement, a template other rural counties can copy.

[Grant] Foundations for Operating the National Artificial Intelligence Research Resource: the NAIRR Operations Center

The National Science Foundation has posted solicitation 25-546 to fund the operations center for the National AI Research Resource, the federal effort to give researchers shared access to compute, data, and models no single campus could buy alone.

NAIRR is the most plausible route to serious compute for tribal colleges, regional public universities, and land-grant researchers working on crop, livestock, and rural health problems. Whoever runs the operations center writes the access rules, so eligible institutions should pull the full solicitation now.

CAL FIRE creates new unit to test emerging wildfire technology

Gov. Gavin Newsom signed SB 1079 creating a Fire Innovation Unit inside CAL FIRE to research, test, and procure wildfire technology, launching formally in January from Sacramento. The agency already runs 1,260 AlertCalifornia mountaintop cameras, 915 of them AI-enabled to spot smoke and alert dispatch before 911 calls, with 200 more due this year.

Those cameras cover 31 million acres of mostly rural private wildland, where detection time decides whether a fire stays small. The same package speeds permitting for fuel reduction and creates a county wildfire preparedness program.