Resources
There is an enormous amount written about AI and almost none of it is written for a 30-person company. This is the short list we actually send to owners — what's worth your evening, what it says, and why we think it matters at your size.
How this list works
Everything here is free or close to it, and nothing on this page is something we sell or get paid to recommend. We add roughly one resource a week and we don't pad the list to hit a number — if a week produces nothing worth your time, we skip it. Items that go stale come off.
The library
Showing all 10 resources.
The federal government's own plain-English primer, and the least breathless thing written on the subject. It lists where AI actually helps a small firm — automating repetitive tasks, drafting marketing copy, handling routine customer questions, finding patterns in your own data — then spends equal time on what to weigh against it: intellectual property, security holes, customer trust. It closes with a glossary of the terms you'll hear in every sales call.
The risk half is the useful half. "Have another person review AI-generated content before it goes out" is the cheapest control a small business can adopt, and almost nobody has actually written it down as a rule.
Mollick teaches this at Wharton, and his argument is that people bounce off AI because they treat it like a search engine instead of a colleague. His prescription isn't a prompt-engineering course: use it as an infinitely patient new coworker, give it context the way you'd brief a person, and spend about ten hours on real work you already know how to judge. The intuition builds itself from there.
This is what we tell owners to do before any engagement starts. You cannot scope an AI project in a business you run if you have never once watched the tool fail at something you're an expert in.
Five self-paced modules — an introduction, using AI for everyday work, prompting, responsible use, and keeping current — that most people finish in under five hours. It runs on Coursera at $49 a month in the US and Canada after a seven-day free trial, so a motivated employee can finish inside the trial week. There's a certificate at the end.
This is the cheapest way to find out who on your team is going to run with this. Buy one seat, offer it around, and pay attention to who actually finishes — that person is your internal owner.
The US standards body's voluntary framework for managing AI risk, organised around four functions: Govern, Map, Measure and Manage. It's deliberately written to apply to organisations of any size and tied to no particular vendor or technology, which is why it has become the reference everyone else points at.
Nobody your size is going to implement the whole framework, and you don't need to. Read the Govern function and steal its questions: who owns this system, who do we call when it's wrong, and how would we even know that it's wrong?
The federal cyber agency's collected guidance on running AI without opening a hole in your business. Most of it is aimed at people building AI systems, but two pieces are written for everyone else: "Deploying AI Systems Securely," for organisations using somebody else's AI, and newer guidance on agentic services — systems that take actions on your behalf rather than just answering questions.
The question underneath all of it is short enough to put in an email, and worth sending to every vendor you're considering: what happens to our data once it leaves our building, and who else can see it?
The FTC's hub for its AI enforcement actions and its guidance to businesses. A lot of it concerns claims — both what AI vendors have been sanctioned for promising and what you're allowed to say about AI in your own marketing. The business blog entries are short and written for non-lawyers.
This cuts two ways for you. Read it as a due-diligence list for the vendors pitching you — and then read it again as a compliance list, because the moment "AI-powered" appears on your own website, these rules are about you.
Stanford's annual eight-chapter survey of the field, free to download along with its underlying data, and the closest thing that exists to a neutral scoreboard. Two figures from the 2024 data are worth carrying around: 78% of organisations reported using AI, up from 55% the year before — and the cost of running a GPT-3.5-level model fell more than 280-fold between November 2022 and October 2024.
That cost curve is the reason "we looked at this a couple of years ago and it didn't pencil out" deserves a second look. The economics of the specific thing you rejected may have moved by two orders of magnitude since you priced it.
A short research spotlight from the SBA's independent advocacy office on the adoption gap between large and small firms. Large firms moved first and moved further; the finding here is that small firms have been closing that gap recently. It's a 610 KB PDF and a quick read.
A reality check that cuts in both directions. You are not as far behind as the headlines imply — and the gap that was protecting you from competitors your own size is closing.
The most-cited field study of AI in a real workplace rather than a lab: 5,179 customer support agents, measured on issues resolved per hour as an AI assistant was rolled out to them in stages. Access raised productivity by 14% on average — and by 34% for the newest and least experienced workers, while barely moving the most experienced ones.
The distribution matters more than the average. At your size the value may be less "make your best person faster" and more "get a new hire to competent in weeks instead of quarters" — which is a different business case, and often a bigger one.
MIT SMR's long-running research initiative on AI and strategy — articles, executive interviews and an annual report, written for people who run organisations rather than people who build software. Some pieces are free to read; the deeper archive is metered.
Read it for the failure patterns, not the success stories. The finding that keeps recurring — that the organisational change is harder and slower than the technology — is the one that decides whether your project lands or quietly dies.
Nothing in that topic yet. We add about one resource a week, so it's worth looking again — or tell us what you were hoping to find at solutions@tellicoai.com.
A note on what's missing
Almost everything written about AI for small business is written by someone selling AI to small business, us included. So the bar for this page is that a resource has to be useful even if you never speak to us. That rules out most of what's published, which is why the list is ten items and not a hundred.
If you've found something that belongs here, send it to solutions@tellicoai.com. We read all of them and we'll credit you if it makes the list.
Bring one process that eats hours every week. Twenty minutes, no pitch — we'll tell you straight whether it's worth solving, and if it isn't, we'll say so.
Book a 20-minute call