Last week Goldman Sachs published a survey of over 1,200 small business owners about their AI usage. The headline number was 75% are using AI in some form. That sounds brilliant until you read the next line. Only 14% have actually embedded AI across their core operations.
I read that and sat with it for a minute, because it describes almost every business I talk to. Three quarters have adopted AI. Fewer than one in five have integrated it into how the business actually runs. That leaves 61% of small businesses in what I would call the dabbling zone. They have got ChatGPT open in a browser tab. They use it to write a social media caption or brainstorm a subject line. Maybe someone on the team tried a transcription tool once. But the processes, the workflows, the way decisions actually get made inside the business, none of that has changed.
The dabbling zone feels productive but it is not
The tricky thing about dabbling is it feels like progress. You are using AI. You can point to examples. Someone wrote a blog post with it last Tuesday. Your ops person used it to clean up some data. If anyone asks, you are an AI-forward company.
But the work still takes the same amount of time. The same bottlenecks exist. The same people are still doing the same repetitive tasks the same way they did 18 months ago. The AI is sitting on top of the old process like a coat of paint on a rotten wall, if that makes any sense.
Less than 25% of the businesses in the Goldman Sachs survey use AI for high-impact tasks. Things like identifying potential customers, optimising supply chains, or generating product insights. The stuff that actually moves revenue. Most of the AI adoption is happening at the edges of the business, not the core.
The three barriers are not what you think
The survey flagged three things holding businesses back. Lack of technical expertise. Difficulty navigating the crowded tools landscape. Data privacy concerns. Over 70% said they would benefit from more training and implementation resources. They know they need to go further. They just do not know how.
And look, I get why that is. Every week there is a new tool, a new model, a new feature drop from one of the big players. It is genuinely overwhelming if you are trying to run a business at the same time. The natural response is to wait until things settle down. The problem is that things are not going to settle down. This is the pace now. Waiting is its own decision, and it has a cost.
What integration actually looks like
When I work inside a client's business, integration does not mean buying more tools. It means picking a real workflow that the team does every week, one that takes too long or involves too many steps, and rebuilding it so AI is part of how it runs. The team does the rebuild with me. They understand why it works. They own it afterwards.
One agency I worked with had their content review process taking something like six separate touchpoints before anything got approved. We rebuilt that workflow in the first three weeks. Same quality standards, fewer steps, and the team freed up roughly 10 hours per person per week. They used that extra capacity to take on two new retainer clients without hiring anyone.
That is what the 14% looks like. Real workflows changed. Real time recovered. Real revenue impact you can measure.
The gap is widening and it will not close on its own
Here is the part that should worry anyone sitting in the 86%. The businesses that have integrated AI into their operations are compounding their advantage every month. They are faster, they are taking on more work with the same team, and they are learning how to spot the next workflow to improve because the first one worked.
The dabblers are still brainstorming subject lines.
If you recognise your business somewhere in this, the honest question is not whether you are using AI. It is whether anything about how your team operates has actually changed because of it. If the answer is no, that is fine, but you should probably stop telling yourself you are keeping up.
I work with owner-led businesses of 5 to 25 people who want to move from dabbling to integration. If that sounds like where you are, here is how the fractional AI engagement works.
