Hands On AI Training

Hands On AI Training Practical AI training for people, local governments, small businesses, and nonprofits.

06/24/2026

If your job buries you in long documents, this one is for you.

Anybody who has served on a board or worked in a city office knows the drill. The packet shows up a few days before the meeting. Ordinances, budgets, staff reports, minutes. Two hundred pages, and somehow you are supposed to know what matters before you sit down.

Here is what is possible now. You load the whole packet into a tool built for long documents and just ask it questions the way you would ask a person. What changed in this budget from last year? Which items need a vote? Give me the third agenda item in five sentences. Google's NotebookLM was made for this, and Claude and ChatGPT both handle big files well too.

The best part is that these tools point you back to the exact pages they pulled from. So you check before you trust it. That step matters. The tool finds it, you confirm it, you decide.

This does not do your homework for you. It makes the homework doable when you have a full-time job and a meeting tonight. A clerk in a small town can prepare the way a big-city staff does, minus the staff.

That has always been the point for me. The capability gap is closing, and it is closing now. handsonaitraining.com

06/23/2026

Here is a small AI habit that saves a lot of time. You already did the hard part. Let AI handle the rest.

Think about the last real thing you wrote. A board update, a grant report, a recap of an event. That single document already contains everything you need to also make a newsletter blurb, a Facebook post, an email to your supporters, and notes for your next meeting. Many people write each of those over again from nothing. You do not have to.

Paste your finished piece into ChatGPT, Claude, or Gemini and ask it to rework the same content into each format you want. Tell it who it is for and how long it should be. You will have drafts in seconds.

Those drafts are a starting point, not the final word. You still read them over, fix the tone, check the facts, and cut anything that does not sound like you. That judgment is yours and it always will be. But editing a draft beats staring at a blank screen at 9 p.m.

If you run a small office with nobody handling communications, this one trick is the difference between getting the word out and letting it slide.

Pick one thing you wrote recently and try it this week. handsonaitraining.com

Most small businesses are using AI now. Very few are using it well. That gap is where I come in.A 2026 Goldman Sachs sur...
06/22/2026

Most small businesses are using AI now. Very few are using it well. That gap is where I come in.

A 2026 Goldman Sachs survey of 1,256 small business owners in its 10,000 Small Businesses program found that 76 percent currently use AI. Only 14 percent have fully integrated it into core operations. And 73 percent said they would benefit from additional training and resources.

I have been saying this since I started Hands On AI Training. Getting access to ChatGPT, Claude, or Gemini is the easy part. They cost less than a phone plan. The hard part is knowing how to use them for your actual work, and almost nobody is being shown that. People open the tool, type something in, get a so-so answer, and quietly decide AI is overhyped.

It is not overhyped. It is untaught.

What I do is simple. I sit down with you, on your computer, with your real work in front of us, and we go until you can do it yourself. No generic slides. No webinar you half-watch. You keep the skill after I am gone.

If your team is using AI but you are not sure it is paying off, let's talk. handsonaitraining.com

Here is something worth doing before summer fully takes over.Six months ago, a lot more people started using Claude, Cha...
06/19/2026

Here is something worth doing before summer fully takes over.

Six months ago, a lot more people started using Claude, ChatGPT, or Gemini for the first time. Some have made it a real habit, learned and refined their workflows. Others drifted back to doing things the old way. But there is a third group that concerns me more than either of those: people who think they have a habit but have actually just gotten comfortable doing the same things the same way, at roughly the same quality they could have managed without the tool.

A midyear checkup has three questions.

Are you using AI for harder things than you were in January? Growth means applying the tools to problems that would have taken much longer six months ago. Getting comfortable looks like the same five prompts, very well-polished.

When did you last push back on an AI output instead of accepting it? The skill is not in getting a draft. It is in knowing when the draft is wrong or generic or missing something only you know. If you cannot remember the last time you sent it back, you may be trusting more and thinking less.

Has AI freed up time you are actually spending on harder work? Efficiency is real. But the point of freeing up time is to use it on judgment calls, relationship work, and decisions that actually require a human. If the time just evaporated, something is off.

These are not trick questions. They are the difference between using AI and getting better at it.

handsonaitraining.com

06/17/2026

Most meeting notes do not survive the walk back to your desk.

You left the room with a sense of what was decided. Your notes are there. But the action items are buried in them somewhere, probably without clear owners or deadlines, competing with every sentence that captured discussion rather than decision.

AI handles this specific problem fast. Paste your meeting notes into Claude, ChatGPT, or Gemini and ask for three things: every action item, who owns it, and any deadline that was stated or implied. Two minutes later you have a clean list instead of a document you meant to review later.

You can also ask it to pull out just the decisions, one sentence each. That is something you can put in the minutes or send to someone who was not in the room. Or ask it to draft a short follow-up summary email: what was decided, what happens next, who does what.

One thing to keep in mind: AI works with the notes you give it. If your notes were thin, it will fill in with inference. Read what it produces before anything goes on the record or into your official minutes.

This works for council meetings, board meetings, staff check-ins, and client calls. The tools are probably already sitting in a browser tab you have open. The skill is knowing what to ask.

handsonaitraining.com

06/15/2026

OpenAI and Visa announced this week that ChatGPT can shop and buy on a user's behalf. Connect your Visa card, tell the agent your constraints, and it evaluates options and completes the purchase. Anthropic and Google are building toward the same capability.

What this means for any small business, nonprofit, or local organization: an AI agent buying on someone's behalf is not browsing your about page. It is not responding to your photography or your brand story. It extracts structured information and either finds what it needs or moves to the next option.

I have watched a version of this play out here in southeastern Minnesota. A daily AI-assisted briefing covers events and local news across the region. When an event listing lives inside a PDF flyer, an image-based social post, or a website that buries the date and location behind multiple clicks, the agent cannot read it. The event disappears from the briefing entirely. The organizer never knows why attendance was soft.

That is not theoretical. The Visa and ChatGPT announcement makes clear it is expanding.

The shift follows the same pattern as search engine optimization in the late 1990s. Businesses that adapted were found. Businesses that did not were invisible. The rules are new. The early movers get the same advantage they got in 1998.

I have a training that walks through exactly this: how to audit your website and online presence from an agent's perspective and fix what is missing. If you want to know where you stand, that is what it covers.

handsonaitraining.com

A small organization's AI policy doesn't need to be thirty pages. Two pages will do more than a binder nobody reads.The ...
06/12/2026

A small organization's AI policy doesn't need to be thirty pages. Two pages will do more than a binder nobody reads.

The question comes up constantly. City clerks, executive directors of small nonprofits, owners of three-person businesses. They hear about state laws and executive orders and big-company frameworks and conclude they need a lawyer and three months to write the thing. They don't.

A workable two-page policy answers five questions in plain English. What's allowed: routine emails, drafting, summarizing, research, brainstorming. What's not: decisions about people, confidential data going into tools that store it, anything that goes out under the organization's name without a human reviewing it. How to handle confidential information: which tools have business-grade privacy, what you never paste in. Who to ask when you're not sure: one name, a board member or director or designated AI lead. And verification expectations, meaning what the human reviewer is responsible for checking before anything ships.

That's the document. Two pages. Plain language so staff actually read it. Reviewed once a year.

You don't need a framework that anticipates every future scenario. You need something today that keeps small mistakes from becoming big ones. The thirty-page version can come later if you grow into it.

This is governance scoped to where you actually are. Not where a Fortune 500 lawyer thinks you should be.

handsonaitraining.com

Anthropic dropped Fable 5 Monday. It is the public version of Mythos, a model Anthropic spent months vetting with select...
06/11/2026

Anthropic dropped Fable 5 Monday. It is the public version of Mythos, a model Anthropic spent months vetting with select industry partners to find and close security vulnerabilities in things like website code. Mythos itself stays restricted. Fable 5 is the version built with additional safeguards for broad public use. It benchmarks above any model that has been publicly available. Genuinely extraordinary.

Here is what most coverage is not talking about: it is available through June 22, then moves to usage credits billed by the token. That is one piece of a larger shift. OpenAI said this spring that unlimited AI plans are "like having an unlimited electricity plan. It just doesn't make sense." Goldman Sachs forecasts a 24-fold increase in token consumption by 2030. The flat-rate era of AI access is ending. Job seekers are starting to ask about their token allocation, not just their salary.

This is the part I want people who are already using AI to hear. You have a real head start. You know how the tools work. That is not nothing.

But there is a difference between knowing how to use AI and knowing how to use it efficiently. If you reach for the most powerful model every time you need to draft an email, you are building a habit that will cost you when pricing changes. Model selection is a skill. Knowing when a less expensive model handles the task equally well is a skill. Knowing when not to use AI at all is a skill.

The people who build those habits now will be in a very different position than the ones who have to change their approach after the meter starts running.

handsonaitraining.com

06/10/2026

A million tokens of working memory. Here's what that actually means.

Anthropic released the latest version of Claude on May 28. The big change for everyday users is the size of what you can put in one conversation. A million tokens is roughly 750,000 words. That's a whole grant solicitation with appendices. A full year of council minutes. A 300-page comprehensive plan. A decade of budgets.

You drop the whole thing in. Then you ask real questions across all of it. What's the trend in our public works overtime over five years? Pull every line about the budget from these minutes. How does this grant in 2024 compare to the 2026 version? Find every place this old policy conflicts with new state guidance.

There's a second thing happening alongside it. Anthropic, OpenAI, and Google are all rolling out agents. Agents don't just answer questions. They read files in your folders, save drafts back to those folders, run on a schedule, and work across apps you already use. Anthropic calls theirs Cowork. ChatGPT is Codex. These models think, and the agent does the work. You still review, edit, and approve.

For a clerk, a director, or a small business owner running lean, this is the part that matters. The capability gap between a one-person office and a department of fifteen is closing in real time. The tools are cheap. Knowing how to use them is the part most people are missing.

That's the training I do.

handsonaitraining.com

06/08/2026

The most dangerous wrong answer from AI is the one that sounds right.

Obvious mistakes are easy. You read it, see the problem, fix it, move on. The hard ones are the answers that look polished, sound reasonable, and include a confident-looking citation that turns out to lead nowhere. Those are the ones that get sent.

A 30-second verification habit handles most of it.

When the AI gives you a number, find the source. If it cites a study, open the link and confirm the quote is there. When it names a person or a title, run a quick search. The fakes look real, and the only way to know which is which is to check.

The rule I use: anything specific you'd repeat out loud, check once before you send it. A grant deadline. A statute. A budget figure. A quote from a state agency. Those are the parts that matter, and they're also the parts AI is most likely to hallucinate.

This is the same standard you'd apply to a smart intern who walked into your office with a draft. You read it carefully. You don't put your name on it until you've checked the load-bearing parts (or at least you should!).

AI gives you a starting point. The judgment about what's true stays with you. That part of the job isn't going anywhere.

handsonaitraining.com

Address

207 Washington Street
Winona, MN
55987

Website

Alerts

Be the first to know and let us send you an email when Hands On AI Training posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

Share

Category