AI advertising agencies: what actually changed, and what didn't
Every advertising agency now calls itself AI-powered. Some of that is real. A lot of it describes tools the ad platforms give everyone for free. If you're comparing AI and advertising agencies, the useful question isn't whether they use AI. It's whether they use it on the parts of advertising that still decide whether you make money.
I've run advertising for forty years, from before the internet to campaigns where the platform writes half the ad. Here's what genuinely changed, what didn't, and how I'd judge an AI advertising agency today.
What AI genuinely changed
Creative variation and testing. An agency can now produce dozens of headline, image and copy variations in the time it once took to make three. That makes real testing affordable for small budgets. The skill has moved from making ads to deciding which ideas are worth testing and reading the results correctly.
Targeting, mostly done by the platforms. Google and Meta now find audiences largely on their own, using your conversion data. Hand-built audience lists matter less. What matters more is feeding the platform the right signal: a tracked enquiry or sale, not a page view.
Automated bidding. The platforms adjust bids constantly. An agency that still sells "bid management" as its main skill is selling something the machine already does.
Follow-up speed. This is the change most owners miss. AI can answer an ad enquiry in minutes, by text, email or an AI agent on the phone, and keep following up until the person decides. The ad gets the click. The follow-up gets the customer.
What AI didn't change
AI can make a hundred versions of a weak offer. It can't make the offer strong. Why someone should choose you, what you're promising, and what you charge are still strategy decisions, and they still decide most of the result.
The landing page didn't change either. If the page doesn't quickly say what you do, why you and what to do next, better targeting just sends more people to leave.
And measurement is still on you. The platforms report their own success. Someone has to connect spend to actual customers and revenue, and say plainly when a campaign should stop.
| Part of advertising | AI's role now | Still needs a person |
|---|---|---|
| Ad creative | Drafts and variations at scale | Which ideas are worth testing |
| Targeting | Largely automated by the platform | Choosing the right conversion signal |
| Bidding | Automated | Setting the budget and the target cost per customer |
| Lead follow-up | Instant replies and ongoing nurture | Hand-off rules and the sales conversation |
| Offer and positioning | Research support | The decision itself |
| Measurement | Faster reporting | Connecting spend to real revenue |
Ads are now showing up inside AI assistants
There's a newer shift too: advertising inside the AI assistants themselves. I wrote about what that means in They sell ads in ChatGPT now. The short version is that paid placement in an AI answer and being recommended in it are different things, and the second still depends on being clearly known, verified and quoted. That's the work of AI search visibility, and it compounds in a way ad spend doesn't.
How to judge an AI advertising agency
Lists of top AI advertising companies won't tell you much about fit, for the same reasons I explain in how to choose an AI marketing agency. Ask instead:
What will you measure: cost per click, or cost per customer? How do you decide which creative to test? What happens to a lead in the first five minutes after it comes in? Who looks at the landing page, and what would you change on mine? When would you tell me to spend less?
An agency that answers the last question honestly is usually worth talking to.
Why follow-up often beats more ad spend
When owners tell me their ads aren't working, the ads are often fine. The enquiries arrive and then wait until the next morning, by which time the person has called someone else. Spending more on ads only produces more enquiries that wait.
Fixing response time is usually cheaper than raising the budget, and it's one of the clearest jobs for an AI marketing automation agency: instant replies, booking, reminders and CRM updates, with a person taking over when it matters. I set that up through AI Automation and AI Agents, and a Done-For-You AI Agent is $3,500 plus $1,000 a month.
How I run AI-assisted advertising
My AI marketing agency runs Google and Meta campaigns against cost per customer, not clicks, with AI-generated variations for testing and conversion tracking set up first. I set the strategy and review the work myself. If you'd rather start with a second opinion on your current ads, a paid strategy hour is $350, and the first 30-minute call is free.
Common questions
What do AI and advertising agencies do differently now?
They use AI to produce and test many ad variations, rely on the platforms' automated targeting and bidding, and answer ad leads faster with automation. The parts that still decide profit, the offer, the landing page and honest measurement, still need experienced people.
What is an AI-powered digital marketing agency?
It's an agency that uses AI throughout its work: research, content, ads, follow-up and measurement. The better ones also help clients get recommended in AI answers from ChatGPT, Gemini and Google's AI Overviews, not just in ads.
What does an AI marketing automation agency do?
It sets up the systems that answer and follow up with leads automatically, including instant replies, nurture messages, booking and CRM updates, with clear rules for when a person takes over. For businesses running ads, it's often the fastest way to get more customers from the same spend.
How do I compare top AI advertising companies?
Ignore the ranking and ask how each one measures success. A good agency reports cost per customer and revenue, explains how it chooses what to test, shows what happens to a lead in the first five minutes, and tells you when to spend less.
About the author. Siamak Kalhor is an AI marketing and LLMO consultant in Los Angeles with more than forty years of experience in marketing strategy. He advises professionals and e-commerce leaders on becoming the answer AI assistants give. Get in touch, or connect on LinkedIn.