The short version: AI stopped suggesting and started doing
The biggest shift in AI over the past two years is not smarter chat. It is that AI moved from suggesting things a human then does, to doing the work itself. The industry word for this is agentic: AI that can take a task, work through the steps, use your actual tools, and hand back a finished result.
For marketing, that changes the economics of a lot of grunt work: pulling reports, checking search terms, tagging leads, drafting variations, moving data between systems. Work that used to eat someone’s Tuesday can now run on its own, with a human checking the output instead of producing it.
The rest of this post covers the three developments that actually matter to a business owner (AI search, AI inside the ad platforms, and agents in the back office) and the parts you can comfortably ignore.
AI search: your next customer may never see page one
People increasingly ask ChatGPT, Perplexity and Google’s AI Overviews the questions they used to type into a search bar, including “who should I hire” and “what’s the best option for X”. The answer arrives as a recommendation, not a list of ten blue links.
That has spawned a discipline alongside SEO: answer engine optimization, making sure that when an AI answers a buying question in your category, it names your business. It rewards much of the same groundwork as SEO, but the win condition is being cited and recommended, not just ranked.
This is the development we would least ignore. Rankings you spent years building can keep existing while quietly receiving fewer clicks, because the answer engine already answered. Our AEO service exists for exactly this shift.
AI inside the ad platforms: more power, fewer dials
Google and Meta have both rebuilt their platforms around machine learning. Google’s Performance Max and AI Max for search decide where ads show and what they say; Meta’s Advantage+ does the same for social. You feed the machine assets, a budget and a goal, and it handles the rest.
The honest trade: these systems are genuinely good at finding buyers, and they take away controls advertisers used to rely on. They are also only as good as the data you feed them. A campaign optimizing toward broken conversion tracking is a very fast way to spend money on the wrong people.
AI ad products amplify whatever you give them. Clean tracking and a clear offer get amplified. So does the opposite.
Agents in the back office: the unglamorous gold
The least hyped development is the most useful one for small and mid-sized businesses: AI agents wired into your actual systems. Not a chatbot on your website, software that reads the enquiry, qualifies it, writes the CRM record, drafts the reply, and flags the ones worth a phone call.
- Lead handling. New enquiry lands, gets enriched and routed, follow-up drafted, before a human has opened their inbox.
- Reporting. Ad spend, website data and sales pulled into one summary that says what changed and why, instead of five dashboards nobody reads.
- Content operations. Drafts, variations, resizing and tagging handled by the machine; judgement and approval kept with a person.
“Agentic” should mean the work gets done, not that there’s a chatbot in the corner. This is the standard we hold our own AI automation work to: fewer hours on busywork, measured against the hours it used to take.
What you can comfortably ignore
Plenty. Most AI announcements are aimed at other AI companies, not at you.
- Model launch drama. Which lab is ahead this quarter does not change your marketing plan. The tools you use will absorb the improvements either way.
- AI-generated everything. Publishing floods of machine-written content is a strategy for getting ignored, by readers and, increasingly, by Google.
- Tools without a job description. If you cannot name the task, the hours it takes today, and what done looks like, you are not buying a tool. You are buying a subscription.
And a when-not-to-spend note: if your offer is unclear or your tracking is broken, AI is not the next investment, fixing those is. Automating a broken process just means being wrong faster.
Where to start this quarter
- Ask an AI about your category. Ask ChatGPT and Google who they would recommend for what you sell, in your area. If you are absent, that is your AEO baseline.
- Audit one repetitive task. Pick the most boring weekly job in your marketing and cost it in hours. That is your first automation candidate.
- Check what your ad platforms’ AI is being fed. Conversion tracking, product data, audience signals. Fix the inputs before judging the outputs.
- Keep humans on judgement. Strategy, offers and anything a customer reads at a decisive moment still deserve a person’s eyes.
None of this requires betting the business on a robot. It requires picking the two or three developments that touch your revenue, and moving on those while competitors are still reading announcements.
Frequently asked questions
How is AI actually used in marketing?
Four main ways: AI search surfaces and recommends businesses when buyers ask questions; ad platforms use machine learning to decide targeting and creative; agents automate back-office work like lead handling and reporting; and generation tools speed up drafts and variations. The first three move revenue; the fourth mostly moves deadlines.
What is the biggest recent AI development for small businesses?
Agentic AI, systems that complete tasks end to end using your real tools, rather than suggesting text for a human to act on. For a small business that means lead follow-up, reporting and data entry can run automatically, which frees the hours that used to go to admin.
Will AI search replace SEO?
It changes what winning looks like rather than replacing the work. Answer engines lean on the same signals as search (clear content, authority, consistent business information) but the goal becomes being cited and recommended in answers, not just ranking. Strong SEO is the foundation; answer engine optimization builds on it.
Should I use AI to write my marketing content?
As a drafting assistant, yes, it is genuinely useful for variations, outlines and first passes. As an unsupervised publisher, no. Mass-produced content reads generic, and Google’s helpful content systems are built to demote it. Keep a human on judgement and anything a customer reads at a decisive moment.
When should a business not invest in AI?
When the fundamentals underneath it are broken. If your offer is unclear, your tracking is unreliable or nobody follows up on leads, AI will only amplify those problems. Fix the process first, then automate it, the other order just gets you to the wrong answer faster.