One-on-one advising for small and medium business decision makers, like a caddie for business tools. Before you buy the software or green-light the AI plan, we talk it through and I do the homework, so you come to the table understanding exactly what you're getting into.
The part of your business where the value gets made, your judgment, your relationships, the work clients actually pay for, no model replaces that and I won't sell you one that claims to. Everything around it is fair game. Ask today's AI to just be smart and it'll sound right while being wrong. Where it earns its keep is using tools, holding context, and following instructions without getting bored: a secretary that actually logs the work, a junior dev for the little tools that keep your operation on the rails. You stay the expert who checks the work, and your week goes back to the parts only you can do.
Find the hours your team burns on copy-paste, and the cheapest, most boring way to get them back. Sometimes that's AI, a lot of the time it isn't.
LLM Usage
What today's models actually do well, where they fall over, and how that maps to your work.
Agents
When an agent beats a script, when it doesn't, and what it takes to run one safely. I run my own daily.
Picking Providers
The AI add-on or the specialty software that's owned this niche for years, cloud or your own hardware, the cheap plan or the expensive one. Which setup deserves your money, with reasons you can check.
Software Audits
A second set of eyes on what you bought, what you were quoted, or what somebody built you.
Build vs. Buy
Custom build, off-the-shelf, or nothing new at all. What each path really costs, before you're committed to one.
Start with one question ★ Sized to the ask ★ Ongoing if it's actually useful
The gaps show up after the invoice.
Here's the usual way it goes wrong: somebody who makes money selling software sells you the thing they sell, not the thing you need. It's the club fitting where the shop only carries one brand, and funny enough the perfect fit is always in stock. It mostly fits, so you sign. The missing parts only surface once you've paid for implementation, and now you're patching around them.
And the sticker price is the small part. It's per-seat billing every month, plus setup, plus everything it takes to move your business onto the new system. Transitions are expensive. AI has an extra wrinkle: usage is cheap right now because investors are subsidizing it, so the real question is what the tool is worth to you when the bill goes real. The cheap insurance is knowing the right questions to ask before you sign.
For what it's worth, I think AI is a bubble. But the dotcom bubble popped and we still got the internet. Some of today's tools vanish with the funding, some become the new plumbing, and telling those apart before you spend is most of the game.
No commission · No stake in what you buy
How it works
This takes a different shape for every business. The loop underneath stays the same.
01
Say the problem out loud.
A short call, no charge, no slide deck. If it's a ten-minute answer, you get it on the spot and we're done.
02
I do the homework.
I dig into the tools, quotes, and workflows in question and test the claims. It usually takes days, not months.
03
You get it in writing.
A plain-language recommendation you can hand to your team, your vendor, or the next person you hire.
04
We shape it from there.
Some businesses need one answer, some want a few weeks of help, some want somebody on call as things keep changing. We'll work out the shape that fits, nothing renews on its own and nothing locks in.
THE USUAL STARTING POINT
Two conversations, one answer.
First meeting is 30 to 60 minutes: you lay out the problem and we jam on it. Then I go away, do the research, and test what needs testing. Second meeting, you get my advice and we talk it through until it makes sense.
This whole page is a test: I'm seeing whether businesses want AI advice from somebody who hasn't already decided AI is the answer. No packages, no retainers, we just talk first.
Who you're hiring.
The job, plainly: I sit on your side of the table. I've got no stake in what you end up buying, so when you sign off on something you know exactly what you're getting into.
The record behind that: I've been building websites for small businesses since 2009, and in a small town the website is never where it ends. You end up in the email setup, the booking system, the client intake, whatever software the industry runs on. For the last decade I've done it as the technical half of Gem Webb, a two-person web and marketing shop in Owen Sound. Seventeen years in the trade makes you quick at learning a new industry and asking the questions that matter. It also teaches you to explain things in whatever language a business speaks. I picked that up even earlier, teaching drum lessons to kids: if the person across from you doesn't get it, that's your fault, not theirs.
And I'm not an AI guy by default, I didn't even like the stuff until late 2025, when the newer models started writing code that actually worked. Now it's the hobby that got out of hand. I run subscriptions, APIs, and local models side by side, so I know what each one is for, what it really costs, and which of the claims hold up. I build and run my own automation and agents every day, so the advice comes from operating this stuff, not from reading about it.
Spec Sheet · B. Campbell
FieldSoftware & AI · Small Business
In the tradeSince 2009
Sites100+ built · 50+ kept running
Day jobPartner, Gem Webb (est. 2015)
SchoolAlgonquin · Fanshawe
AIAgents in production, daily
StackSubscriptions · APIs · local models
Off hoursOn the course
BaseLondon, Ontario
Interactive Multimedia (Algonquin) · Web Development (Fanshawe) · The rest learned by shipping
A good fit if…
You run a small or mid-sized business and think AI or automation could probably help, you're just not sure where to start.
You're holding a software quote or an "AI strategy" pitch and want a second opinion before you sign.
You're the one who has to sign off on a tech decision, and you'd rather walk in understanding it than just trusting the vendor.
Repetitive work is eating hours and you want the reliable fix, not the impressive one.
You'd like a coach on call: somebody who follows this stuff for fun and flags when something new actually matters for your business.
Not a fit if…
You need a team to build and maintain a big custom application. That's a build crew, not an advisor.
You want someone to resell you licenses or take a margin on tools. I don't.
You want "AI" in the press release more than in the workflow.
Are we a fit?
A couple of questions, sixty seconds. If I'm not your guy, you'll leave with directions to whoever is.
THE FIT CHECK · No email required
What do you need right now?
THE FREE VERSION OF THE ANSWER
You're closer than you think.
You don't need an AI strategy. The annoying part of your week that you already complain about, that's the starting point, and knowing it puts you ahead of most demos. And don't rule out the boring specialty software that's owned your industry's problem for decades. AI being stuffed into everything doesn't make it the answer to everything.
Want to get your bearings first? AI is a self-starter's game, so skip the reading list: get a basic Claude plan (the twenty bucks beats camping on the free tier) and have it build you a small tool for something you know inside out, your business or better yet a hobby. You'll see where it's smart in an afternoon, and in a domain where you're the expert you'll catch it being confidently wrong. When you'd rather talk it through with a person, that's what the form below is for.
THE FREE VERSION OF THE ANSWER
It's okay to slow this down.
A real vendor survives a two-week pause. While it's paused, three fair questions: what does month twelve cost all-in, what happens to your data if you leave, and can they demo the exact feature you're buying on your data instead of theirs. Honest vendors answer all three without flinching.
If you'd rather have somebody independent read the fine print with you, somebody with no commission riding on your yes, that's exactly the job.
THE FREE VERSION OF THE ANSWER
The most fixable problem on the list.
Good news: repetitive work is where fixes are cheapest and pay back fastest. Before buying anything new, search the name of the software you already pay for plus the word "automation". Half the time the feature already exists and just isn't turned on. A rough count of the hours the chore eats each week tells you what a fix is worth to you.
Want a second brain on which fix actually suits your operation? Bring that hours count to the form below and we'll start there.
THE FREE VERSION OF THE ANSWER
Judge them on your work, not their demo.
You're allowed to try before you commit. Most serious vendors have a free tier or a trial month, so run a few real tasks from your actual week through each one and watch for consistency, not the one impressive answer. Reviews from businesses your size tell you more than the big-logo case studies.
If you'd rather have the bake-off run properly and the answer in writing, that's the job.
Not yet, and that's honestly fine.
Paid advice only makes sense once the problem costs you real money. And AI is a self-starter's game, nobody can hand it to you. The best twenty bucks you can spend: a basic Claude plan and an afternoon having it build you something real, a little tool or a simple page for your business or a hobby you know inside out. Watching it build is where you'll see AI being smart. Using it where you're the expert is where you'll catch it sounding right while being wrong.
For the classroom side: YouTube is the lowest barrier to entry, you get to watch things happen as somebody actually does them. Blogs and newsletters are better if you'd rather take your time and move at your own pace. Either way, browse until you find somebody who shows their work instead of hyping it. I've got favourites, but this field moves too fast to carve names into a website, ask me and I'll tell you who I'm watching and reading this month.
And if you're a self-starter who asks good questions, the LLM itself can be a great teacher: have it walk you through what to type, what files to make, and how to organize things. Point Claude at the documentation for whatever you're working in, or talk through what your business is trying to do, how the popular players pull it off, and whether you could replicate it. Fair warning, LLMs are confidently wrong often enough that this route is for folks who trust their own self-guided learning, or who already know the domain well enough to catch it.
Prefer something structured? Coursera has been a great platform for over a decade, and the industry's own beginner courses live there: Google's AI Essentials and the classic AI for Everyone.
Then keep an eye on the hours the problem eats. When it turns into real money, come back.
THE FREE VERSION OF THE ANSWER
You don't need another newsletter.
You need a filter with your name on it. That's the coaching side of this: I learn your business and your stack, I keep up with this stuff because I honestly enjoy it, and when something ships that actually matters for you, you hear about it from me with the why-it-matters-to-you attached. Everything else, you never have to see.
Not sure that's worth paying for? Fair. There are solid free newsletters that do the general version of this. The difference is that none of them know your business, and I would.
You need a build crew, not an advisor.
Building and maintaining custom software is a team sport. Referrals from other business owners beat everything: ask who built their stuff, whether they'd hire them again, and ask to talk to a client the shop has kept for more than a year.
The part that is my lane: sanity-checking their quotes and plans before you sign anything.
That's my day job, different hat.
Websites, marketing, and ads for small business is what we do at Gem Webb, and have for a decade. Head over there. You'll still get me, just wearing the other hat.
Knowing the term is half the search. Look up "managed IT provider" plus your town, then ask two questions: what's the guaranteed response time, and how do they test their backups. Anyone who dodges either one is out.
If you end up torn between two of them, that second opinion is something I do.
You don't need a consultant, honest.
AI is a self-starter's game, and the fastest lesson costs about twenty bucks: a basic Claude plan (or Cursor, if you're up for its tech curve) and a weekend having it build you something for a hobby you actually know. Simple pages and little tools, Claude calls them artifacts, are the easiest way to see where AI is smart. Using it where you're the expert is how you see where it's dumb: it misses context, and it can sound sure of an answer it doesn't actually have.
Ready to go deeper? YouTube is the classroom, browse until you find a teacher at your own speed. For structured learning, Coursera's AI courses are the real deal. And Grok isn't free, but it's impressed me.
Tell me the problem.
In your own words, plain is better. What's slow, what's confusing, what somebody's trying to sell you, what you think a robot should be doing by now.
The fix might be a dashboard, a bot, an agent, a piece of specialty software, or honestly just a person. Figuring out which is most of the job.
I read every one, and "you don't need me for this" is a real answer you might get.
Real reply within two business days ★ If I'm not the right fit, I'll say so
Not sure you even have a question?
That's usually the best starting point. Send the half-formed version, figuring out what to ask is part of the job.
