Grass-Roots AI for Washakie County

Ten thousand dollars. One award, one time, from the Wyoming Business Council’s Community Economic Growth Grant1 — a program that exists to fund projects improving a community’s quality of life through economic development, and whose ceiling is exactly $10,000.

It pays for a twelve-month pilot program in Washakie County: twelve working websites built for local institutions and small businesses, four public workshops, and a written method anybody can follow afterward without paying anyone.

That is the whole ask. This grant has no second phase, no ongoing subsidy, and nothing to renew.

Describe what you want, in your own words. That is the whole of your part.

The way you would say it to somebody across a counter: I run a tree service. People want to know if I am insured and how fast I can come. I work the Basin from Greybull to Thermopolis. That is enough to start.

Your AI does the rest — the writing, the building, checking every fact against a source, making sure it reads on a phone and prints on paper. You read what comes back and say what is wrong with it; it is fixed, and you say what is wrong with that. If you can tell when a description of your own business is wrong, which you certainly can, you can direct it.

Because the state’s artificial-intelligence money has an address, and it is not here.

In January 2026 the Laramie County Commission approved a 1.8 gigawatt data center outside Cheyenne — six hundred acres of building and another six hundred and fifty-nine for the power plant that feeds it, four hundred permanent jobs when it is finished2. In June, at its customer’s request, the developer paused work on it, with no date to resume20 — which is not an argument against the project and was never the point. The point is that a bet of that size can go quiet in five months, and a county of this size has no way to plan around it either way. The University of Wyoming runs the state’s AI initiative from Laramie, with roughly four million dollars of federal money for computing hardware3.

In June 2026 the Governor issued an executive order setting the state’s terms for this industry — water, wildlife, utility bills, benefits for the local governments that host a facility, and training for careers in digital infrastructure through the university, community colleges and school districts9. A month earlier the Legislature’s Select Water Committee heard testimony on how much water and power data centers draw8. Every one of those terms is about data centers themselves, which is the right concern for a gigawatt.

All of it is real, and none of it is a complaint. But a county of about 7,600 people, some nine hundred fewer than at the 2010 census4, does not get a data center and does not get a research center. It gets whatever it builds for itself.

Wyoming is building the engine room and planning the workforce to staff it. Nothing in the order teaches the rest of the state to drive.

Five words are used throughout this page. They are worth fixing before anything else.

• The sponsor — the local body that holds the grant and receives the deliverables, since an individual cannot be the applicant. Not yet found; it is the first thing to do.
• A subject — the office, business, club, trade or question a site is about, and the person who speaks for it. Twelve of them. A subject chooses a name and hands over the material, signs up for nothing and holds no password. At the end of the pilot program the site is handed across and the subject becomes its owner, domain and all.
• The trainees — local people who come to the four workshops and learn to keep a site: to change the words, publish the change, and check the sources behind it. A few of them are the whole of the plan for what happens after the year.
• The ambassadors — hoped for rather than arranged: the county’s FFA chapters, who would carry pages like these to the farms. Nobody has asked them yet.
• The trainer — whoever runs the workshops. Here that is the person doing the work; when he is building a site rather than teaching, the same person is simply the builder.

Twelve websites, four workshops, and a method left behind in public.

• Twelve one-topic sites for Washakie County subjects, such as a county office, the library, the fair board, a rancher, a welder, a bookkeeper, whoever comes forward. One topic each, built to the same format. High school students might have their first-ever AI project shown to the world.
• Four public workshops, open to anyone, where the work is done in the room rather than described. They are also where the county’s own help comes from: a few local people given a light introduction to AI, who can afterward make a change for a neighbor’s site.
• A written method, published free, so the thirteenth site costs nobody anything.
• One AI subscription, the Claude plan the sites are built with, for the twelve months of the pilot program and no longer12.

Each page is one HTML file — no framework, no build step, nothing to install — and a site is that page plus a few small things beside it: an icon, a social card, a list of its own links. There is no database and no server. The hosting costs nothing at all; the only cost is the name, roughly $11 to $28 a year in domain registration depending on the ending, and nothing else7, and the person it belongs to owns it outright.

Choose a name. That is the whole of it.

The domain, the hosting, the security certificate and the DNS are registered and run from the pilot program’s own accounts. A subject signs up for nothing, installs nothing, and holds no password. What they supply is the material — the ordinance, the price list, the certifications, the record of work — and a name they would like, taken from the .online family so that availability is never the obstacle.

The sheet a site is set up from — page D — can be seen here with every field switched off10. It is shown to make the point concrete: that is the page a subject never has to see.

The only recurring cost is that domain, bought at cost and paid two years forward out of this grant: for a .online name, under $5 for the first year and about $28 a year after that7. At the end of the pilot program the site is handed over — the domain moved into the subject’s own registrar account if they want it, the folder handed across if they do not — and after that the renewal is theirs to keep or to let go.

Or no name of their own at all. A group can share one: a farm cooperative buys washakiefarms.online once, and every member’s site lives under it free, at an address like eastbench.washakiefarms.online, with one index page listing them all. It is like license plates — everyone gets one, a vanity plate costs a little extra each year, and it is up to you.

Every GYR-O page comes in two parts: Intro and Info. Intro, above the double line, is whatever this subject needs — its name, a picture, a paragraph — and it differs from site to site. Info, below the line, is the standard layout, the same on every GYR-O site: three columns of plain answers, sources and technical detail. A reader who has used one already knows where to look on the next.

One topic, laid out to serve two readers at once: the newcomer who knows nothing about it, and the reader who already knows the subject well. Plain answers on the left assume nothing. Full technical detail on the right assumes a great deal. Neither has to be watered down for the other, and neither has to wade through the other. On a phone the columns become buttons; printed, they become one document.

The second idea is about footnotes, and it is overdue. Printed notes at the foot of the page go back at least to the Bishops’ Bible of 156821, and the way a reader reaches them has hardly changed since: leave the sentence, find the mark, come back if you can. Instead of being buried at the bottom of a page, tucked behind a link, or scattered wherever they happen to fall, every reference is gathered into one column of its own — centered in the layout, given the same visual weight as the writing, and always in the same place. Each entry says what the source is, who made it, what you would find there, and what it is doing on this page. A citation is a glance away rather than a journey, and often complete enough that you need not go to the original at all.

The third advantage is money, and it is the one that makes a pilot program like this possible at all. Operating expense is close to zero. There is no server, no database, no content system, no subscription and no vendor — the hosting is free, the security certificate is free, the typefaces are free, and the only recurring cost is the domain name at about a dollar a month7. A site that ten years ago would have needed a content system, a hosting contract and somebody to maintain all of it is now a single page that current web technology serves for nothing, and that AI makes quick enough to write in an afternoon rather than two weeks.

You are reading one now. This page is the proposal and the sample.

There is remarkably little to worry about, because there is almost nothing to attack. No logins, so no passwords to steal. No database, so no records to leak. No forms and no code running on a server, so nothing to break into. No cookies or tracking, so no visitor data to lose. A site is a few plain files handed out to whoever asks, and the risk to the county is remarkably low.

The one thing left to guard is the pilot program’s own publishing account, which is why a subject holds no password at all. The full reasoning is set out in the format’s own reference6.

worland-dog-law.online5 is a finished instance, built before any grant money existed. It takes the Worland municipal dog ordinances, sets them out in plain language, links every claim to the ordinance text, and adds an audit of where the published code and the real code disagree.

It was built for the cost of a domain name. It is the argument for the whole pilot program: this is not a proposal to find out whether the idea works.

A second instance is running on an entirely different subject: baconian-ciphers.online11, a dated research record kept in the same three columns and not revised after publication.

And two samples of what the pilot program would make. A page for a Washakie County farm13 — an invented farm, with every fact about the county, the crops and the rules real and sourced. And a page for a Washakie County tree service15 — an invented company, whose first answer is its insurance, with every line of it linking to a check somebody else keeps. Both are under Other sites.

Walter Wilkinson, of Worland. Retired technology engineer, systems engineering degree from Boston University, 20+ years in the Boston software industry before Wyoming.

The format6 and the working instance were both built here in Worland, unpaid, before this application existed.

Two groups, and neither of them is me.

The trainees. A few local people, trained in the workshops, who can make a change on somebody else’s site. That is what the four workshops are for, and it is the part of the pilot program that decides whether any of it survives the year.

The ambassadors, if they are willing: the Future Farmers of America. The county has two chapters — Chief Washakie FFA, at Worland High School, and Ten Sleep FFA, both rated Gold in 202614. They are already in the farms and ranches this format is built to serve, they are already organized, with advisors and a calendar, and their members would be publishing work that anyone outside the building can see. A page kept by a student for a neighbor’s operation is this whole pilot program in one sentence.

Nobody has asked them. Not a phone call, not an email, not once. It is written down here before it is acted on, so that it can be checked later against what actually happened. If the chapters are not interested, the trainees are still the answer, and nothing in the pilot program depends on the FFA saying yes.

As this page stands, yes — and that is a weakness, not a plan. One funder with one deadline is a single point of failure for a proposal whose other single point of failure is finding a sponsor. Two have now been looked at properly.

• USDA Rural Business Development Grants are the real second route18. The eligible applicant is the same shape as here — a public body, a tribe, or a nonprofit serving a rural area — so the work of finding a sponsor is not wasted if the first application fails. Eligible uses name “training and technical assistance… business counseling and training,” which is what four public workshops and twelve built sites are. There is no cost-sharing requirement, and, unusually, smaller requests are given higher priority, which makes a $10,000 ask a strength rather than an apology. Worland is far below the 50,000 population ceiling. The catch is timing: the 2026 window closed on 30 June, so the next is a summer deadline — after the Business Council’s 1 March, not alongside it. That makes it the fallback, in the right order.
• Wyoming Humanities was checked and does not fit19. It funds registered nonprofits only, requires a humanities scholar on the project and a one-to-one match, and its own grants page says it currently has no state appropriation and reduced federal funding. Listing it anyway would pad the page. It is named here so that nobody spends a week discovering this independently.

Not yet examined: the Wyoming Community Foundation, the county and city’s own discretionary and sales-tax funds, and the school district’s own channels if the district ends up the sponsor. Those are named as unfinished work rather than claimed as options.

The work continues, slower and smaller. The format is already built, already published and already free. Nothing here depends on the money.

What the money buys is pace and reach — twelve subjects in a year instead of two, workshops instead of word of mouth, and a documented method instead of one person’s habit. That is a fair thing for a grant to buy, and an honest thing to say out loud.

Training the workforce, especially young people. The pilot program proves the method on twelve real sites. One of the most important goals of the later stages is to teach people here to do the work themselves — above all the young. The state is planning a workforce to staff its data centers9; this is the other half: people in the county who can put AI to work where they live.

Those later stages are not part of this request. The $10,000 pays for the pilot program alone; anything after it would be proposed separately, on the strength of what the pilot program shows.

Two, stated plainly.

Somebody has to apply. The program funds communities and organizations, not individuals, so a local body has to be the applicant — the city, the county, the library, the school district, the Chamber of Commerce16, or an economic development organization. Finding that sponsor is the first task, not an afterthought.

Twelve subjects have to want one. A site nobody asked for is a site nobody maintains. The pilot program is built around subjects who come forward, and the number is a target rather than a promise.