I’ve researched it, and over $547 billion in AI investments failed to actually do what they were supposed to in 2025.
The technology works. The way people use it doesn’t.
Whilst everyone chases the latest shiny AI thing, more than 80% of AI projects fail…
Twice the failure rate of normal IT projects. For small businesses, it’s even worse.
The problem isn’t the technology. The problem is how we’re thinking about it.
The Real Question Nobody’s Asking
When I talk to local service business owners about AI, they’re drowning in promises about how it’ll either save the world or destroy it. Revolutionary breakthroughs. Robots taking over. Total game-changers.
But here’s what actually matters: Who benefits? Who gets harmed? Do the tools work for the people using them?
These questions get buried under hype.
Yet they’re the only questions that determine whether your AI investment succeeds or joins the 80% failure pile.
I’ve spent nearly a decade working with allied health clinic owners, professional services firms, and home service businesses. I’ve seen what works and what’s a waste of money. The pattern is clear.
Complex AI is failing small businesses.
The businesses succeeding with AI aren’t using the fanciest models or the most advanced systems.
They’re using boring, practical tools that solve specific problems.
Why Using Different AI Tools Actually Makes Sense
Forget the idea that you need one massive AI model to handle everything.
The companies getting real results use a mix of AI tools. Big ones for tricky stuff. Medium ones for everyday tasks. Small ones for specific, repetitive jobs.
Each model handles what it does best.
Airbnb built their customer service AI on 13 different models, trained on tens of thousands of real conversations.
The result?
A 15% reduction in contacts requiring human agents whilst improving satisfaction scores.
Meta uses a similar approach for advertising.
Different models handle different aspects of ad targeting, creative optimisation, and performance prediction.
This isn’t revolutionary. It’s practical.
For local service businesses, this means you don’t need some massive fancy system. You need the right tool for each job.
A small model can handle appointment booking. A medium one can sort through leads. A bigger one can handle complicated customer questions.
This approach works because you’re matching the tool to the job.
You’re not overpaying for computing grunt you don’t need. You’re not coming up short because your tool can’t handle the complexity.
Making Your Team Better, Not Replacing Them
Computer scientist Ben Shneiderman argues that technology should make humans better at their jobs, not replace them.
I’ve seen this play out repeatedly with small business owners. The AI that works builds confidence…
The feeling that comes from knowing you can actually get stuff done.
When your AI chatbot captures leads at 2am, you’re not being replaced. You’re being enhanced. You wake up to people ready to book, rather than missed opportunities.
When your AI system handles the same boring customer questions over and over, your team focuses on complicated problems that actually need a human brain. They’re not doing less work. They’re doing more valuable work.
The goal isn’t to make technology seem human.
The goal is to make humans more capable.
I’ve watched business owners try to automate everything, then wonder why customers feel disconnected.
I’ve also watched them automate the right things and suddenly have time to build deeper relationships with their best clients.
The difference?
Understanding what should be automated and what shouldn’t.
Good AI tools build confidence. They handle tasks you know need doing but don’t have time for. They free you to focus on work that requires your expertise, your judgement, your human touch.
Bad AI tools create dependency. They promise to replace you, then fail at the nuanced work that actually matters.
The Data That Matters
Let me share what’s actually happening out there.
56% of small businesses now use AI, with 63% using it for marketing. 94% of owners project growth for 2026. The adoption is real.
But here’s the problem: only 29% of small businesses keep an eye on their AI systems. Just 36% have someone in charge of them. Over 90% have absolutely no idea what their AI is doing once it’s up and running.
This explains the failure rate.
You can’t succeed with tools you don’t understand. You can’t improve systems you don’t monitor. You can’t fix problems you can’t see.
The businesses I work with that get results from AI share common traits.
They start with clear numbers to track. They watch how things are going. They adjust based on what they see.
They treat AI like any other business system…
Something that needs management, optimisation, and ongoing attention.
Small businesses lose leads to competitors using AI because response time matters.
Manual response converts leads at 40-50%. AI response systems convert at 60-70%. That 20-point difference adds up fast.
But you only capture that advantage if you’re actually measuring it.
What Actually Works
I’ve implemented AI systems for local service businesses across multiple industries. The successful implementations share patterns.
They start small.
One specific problem.
One clear number to track.
One result you can measure.
A physiotherapy clinic wanted to reduce no-shows. We built an AI system that sends personalised appointment reminders, handles rescheduling requests, and confirms attendance. No-shows dropped 40%. That’s boring AI that works.
A home services company wanted to capture after-hours leads. We implemented a chatbot that qualifies enquiries, books appointments, and routes urgent requests to on-call staff. They capture 40% more leads without hiring additional staff. That’s boring AI that works.
A consulting firm wanted to reduce time spent on proposal writing. We built an AI system that generates first drafts based on client requirements and past successful proposals. Proposal time dropped 60%. That’s boring AI that works.
None of these solutions are revolutionary.
They’re practical applications of available technology to specific business problems.
The return on investment is immediate and you can measure it. Running costs are $200-$600 monthly. No massive fancy system required. No technical know-how needed from business owners.
The Window’s Closing
AI customer service will become standard within 12 to 18 months.
Right now, early adopters get ahead of the pack. They respond faster. They capture more leads. They run more efficiently.
But that edge disappears once everyone’s doing it. Then it just becomes what customers expect.
The businesses that get AI right now will dominate their markets. The businesses that wait will spend the next few years scrambling to catch up.
I’m not saying this to create urgency. I’m saying it because I’ve watched this pattern repeat across every technology shift I’ve seen in nearly a decade of working with small businesses.
The window for competitive advantage is real. It’s also temporary.
What This Means for You
You don’t need revolutionary AI. You need boring AI that solves real problems.
You don’t need the fanciest models. You need the right tool for each task.
You don’t need to replace humans. You need to enhance human capability.
You don’t need massive budgets. You need clear metrics and measurable outcomes.
The businesses succeeding with AI aren’t chasing hype…
They’re putting in place practical fixes to real problems. They’re tracking results. They’re adjusting based on what the numbers tell them.
They’re treating AI like any other business system…
Something that needs strategy, implementation, and ongoing management.
That’s the case for boring AI. Tools that work for you, not promises that burn your cash.
The technology’s ready. The question is whether you’re ready to use it properly.
