· AI
The AI Automation Question Isn't Whether to Start. Your Staff Already Did.
The AI automation project most businesses are missing is not a new one to launch. It is the one already running on someone's personal laptop, unmanaged and unmeasured.
A small business owner asking whether to "start" AI automation is usually a step behind the actual state of their company. PagerDuty's 2026 workplace survey of 1,250 office professionals at companies with at least $500 million in revenue found that two out of three have used an AI tool at work that they believed was not permitted. Of that group, 88 percent had shared work-related information with a public AI tool, and 31 percent had shared financial or confidential documents. This survey polled larger companies, but the dynamic it captures, individuals adopting a free or personal tool faster than any policy can catch up, does not require a large IT department to happen. If anything, a smaller company with no formal AI policy at all has less friction slowing it down, not more.
The steelman for locking it down first
There is a real case here, and it deserves to be stated plainly before I argue past it. Handing your customer data or financial records to a public chatbot is not a hypothetical risk, it already happened at scale according to that same survey, and a business owner who reacts by wanting rules in place before anything gets "official" is being responsible, not paranoid. There is also a real case for caution about formal AI projects generally. According to CIO Dive's reporting on S&P Global Market Intelligence's 2025 Voice of the Enterprise survey of more than 1,000 IT and business leaders, 42 percent of companies abandoned most of their AI initiatives in 2025, up from 17 percent the year before, and the average company scrapped 46 percent of its AI proofs of concept before they reached production, citing cost, data privacy, and security as the top obstacles. If you take that data at face value, a small business owner could reasonably conclude the safe move is to do nothing formal until the failure rate improves.
Where that reasoning breaks
Look closely at what actually failed in the S&P Global data: initiatives that companies deliberately launched, budgeted, and staffed, top down. That is a different animal from what PagerDuty measured, which is employees finding a tool on their own and getting real, if unmeasured, value out of it fast enough that more than half who got caught kept doing it anyway despite informal or formal pushback. One of these categories is failing at a rising rate. The other is already working, for free, without anyone approving a budget. Concluding "AI initiatives are risky, so wait" from that pair of facts gets the lesson backward. The formal, expensive, top-down version is the one with the bad track record. The bottom-up version already cleared its own market test, it just happened without anyone deciding it should.
The honest reading, and this is my opinion, not a figure from either survey, is that most small businesses do not have an AI adoption problem. They have a visibility problem. Someone in accounts payable is already using a chatbot to draft vendor emails faster. Someone in sales is already summarizing call notes with a free tool. Nobody wrote that down, measured the hours saved, or checked what data left the building to make it happen. That is not a business without AI automation. It is a business running an unaudited pilot with unknown security exposure and no name on it.
What to actually do this week
Skip the RFP for an automation platform. Start with a five-minute, no-blame conversation: ask three or four people directly what AI tool they already reach for and what task it replaces. PagerDuty's own numbers explain why you have to make it genuinely safe to answer honestly, since roughly half the people caught using an unsanctioned tool in that survey faced some consequence for it, which is exactly the incentive that pushes the behavior further underground instead of ending it. Once you know what is already happening, apply the same discipline to each one: how often does it run, can you put a real number of hours against it, and is the data involved something you would be comfortable explaining to a customer if it leaked. Whatever clears that bar is worth formalizing with a sanctioned tool that does the same job with actual data controls, so the convenient shadow version has nothing left to offer. Whatever does not clear it can stay informal, because banning it outright without a substitute just re-creates the same underground demand you started with.
This is the exact audit MojoAI runs before recommending anything: find what is already running, measure it honestly, and decide what earns a real rollout. If you suspect your team is further along on AI than you are, that conversation is worth having before a compliance question forces it.
Sources
References used in this article. Links also appear alongside the relevant claims.
Tell us what's on your mind.
You don't need a polished brief to reach out. A two-line email about what's bugging you is plenty; we'll tell you straight if we're the right fit, and what we'd tackle first.
We'll scope the work around your workflow, goals, and timeline before quoting anything, so you know what's included before committing.