AI Is Reshaping Digital Marketing Right Now. Here's What We've Seen.
Think back to February 2020. If you were paying close attention you might have noticed a few people talking about a virus spreading overseas, but most of us were busy getting on with life. The economy was moving along, your kids were in school, you were planning holidays and shaking hands without a second thought. If someone had told you they were stockpiling toilet paper you would have thought they'd completely lost the plot.
Then over about three weeks, everything changed. Your office closed, the kids came home, and life rearranged itself into something you genuinely wouldn't have believed a month earlier. The speed of it caught everyone off guard.
I believe something of similar magnitude is happening right now with AI, and the reason I'm writing this is because most of the people I speak to haven't yet grasped how far it's already come. Not in a "one day this will be interesting" way, but in a "this has already changed how we work every single day" kind of way. The people working in this space aren't making predictions about the future. They're describing what has already happened to them.
How We Got Here
At Coffee Marketing Digital we've lived through this transformation firsthand, and honestly the speed of it still catches me off guard. In mid-2025 we joined the AI Inner Circle, a global community of digital marketing professionals led by Mike Rhodes, one of the most respected voices in Google Ads. The premise was straightforward enough: learn how to use AI to build better tools, better reporting and better outcomes for clients. What happened next was anything but straightforward.
When we started, AI was a useful assistant. You could ask it to draft an email, explain a concept or help you structure a spreadsheet. Genuinely handy, but clearly a tool that needed you to hold its hand through most things. Then the models got better. And then better again. Each new release wasn't a small incremental improvement, it was a step change that made the version from a few months earlier feel like something from a completely different era.
Here's what this looked like in practice for us. We set out to build a client reporting and analytics platform called the Coffee Marketing Analyser. The idea was a custom React application connecting to BigQuery, pulling real Google Ads data and presenting it in ways that actually help eCommerce businesses make better decisions about where to spend their money and how to improve profitability. A traditional development agency would quote you months of work and tens of thousands of pounds for something like that. Our three-person team, myself, Ross and Carrie, has shipped 65 versioned releases of this platform. It now includes basket-level conversion analysis, search intent classification, automated market intelligence reports, competitive analysis tools, executive performance reports, break-even calculators and a full PDF report generation system. We built all of it not by writing every line of code manually, but by describing what we needed in plain English and working alongside AI to make it real.
Just this morning I described a new feature I wanted. By the afternoon it was built, tested and deployed. Not a rough prototype either, the finished thing, ready for clients. That is the speed at which this is moving.
The Models Keep Getting Better, Faster
For years AI improved at a pace you could keep up with. Big jumps occasionally, but spaced out enough that you had time to absorb what had changed. Then in 2025 new techniques for building these models unlocked something different. The pace of progress itself accelerated, and it hasn't slowed down since.
To put some context around this: in 2022 AI couldn't reliably do basic arithmetic. By 2023 it could pass professional examinations. By 2024 it was writing working software and explaining graduate-level science to people who actually understood the subject. By late 2025 some of the best software engineers in the world had handed over the majority of their coding work to AI systems and were focusing their own time on higher-level architecture and problem solving. Then in February 2026, new models arrived from both OpenAI and Anthropic that made everything before them feel like ancient history.
There's an organisation called METR that tracks this with proper data. They measure how long a task can be, in terms of how long it takes a human expert, before AI can complete it end-to-end without any help. A year ago the answer was roughly ten minutes. Then it was an hour. Then several hours. The most recent measurement showed AI independently completing tasks that take a skilled human nearly five hours, and that number is doubling approximately every seven months. If you extend the trend line, we're looking at AI that can work independently for days within the next year... and weeks within two.
"But I Tried It and It Wasn't That Good"
I hear this constantly. And I get it, because for a while it was true. If you tried ChatGPT in 2023 or early 2024 and came away thinking "this makes things up" or "I don't see what all the fuss is about", you were absolutely right. Those early versions were genuinely limited and frustrating to work with.
But that was two years ago, and in AI terms two years is ancient history. The models available today are unrecognisable from what existed even six months back. The debate about whether AI is "really getting better" has been conclusively settled by anyone who has actually used the current generation of tools. If you're still making that argument, it's most likely because you haven't tried the latest models or you're basing your opinion on an experience that simply isn't relevant anymore.
Part of the problem is that most people use the free version of these tools, which is typically over a year behind what paying users have access to. Judging AI based on free-tier ChatGPT is a bit like evaluating what smartphones can do by pulling out a flip phone from 2005. The gap between free and paid is enormous right now, and it's growing.
What This Actually Means for Your Business
The AI labs made a deliberate strategic choice early on. They focused on making AI excellent at writing code first, because building better AI requires writing a lot of code. If AI can write that code, it accelerates the process of building the next, smarter version of itself. That strategy worked, and now they're applying the same capabilities to everything else.
This is directly relevant to every eCommerce business and every marketing team in the country. At the end of the day, it's about outcomes, and the outcomes we're seeing from applying AI to real marketing problems are genuinely transformative. Here's what that looks like in practice:
Campaign management and optimisation. AI can already analyse campaign structures, identify wasted spend, classify search terms by commercial intent and generate ad copy at a level that rivals experienced PPC managers. We've built tools that do exactly this. Our search intent classifier processes thousands of search terms and categorises them by buying intent in minutes, work that used to take a senior analyst the best part of a day. The results are specific, clear and actionable, which means the recommendations actually get implemented rather than sitting in a spreadsheet.
Reporting and analysis. We built an entire analytics platform that connects to live Google Ads data, calculates profit-based metrics rather than just revenue, and generates executive reports automatically. The AI didn't just help with bits of this. It built the dashboards, the data pipelines, the PDF generation system and the deployment infrastructure. The whole thing, from concept to working product.
Market intelligence. Our automated market intelligence system monitors industry trends, competitor activity and consumer sentiment across every client's sector, and it runs daily. A year ago this kind of competitive analysis was a quarterly exercise that required days of manual research, if it happened at all. Now it's part of the daily workflow.
Client communication. Automated daily briefings, meeting note processing, structured email drafts based on real account data. The operational overhead of running a marketing agency has been fundamentally reduced, which means more time spent on actual strategy and less on admin.
The Pace Is the Point
What catches most people off guard isn't what AI can do right now. It's how quickly the capability is expanding. The tools we built six months ago have already been rebuilt and improved because the underlying models moved on. The workflows we established in September are now obsolete. The models that genuinely impressed us in November feel limited compared to what's available today.
Think about that for a moment. If AI shows even a hint of a capability today, the next generation will be properly good at it. This isn't linear improvement where things get a little bit better each year. Each generation of AI builds the next, which is even smarter and then the next... exponentially. Dario Amodei, CEO of Anthropic, the company behind Claude, has publicly stated that AI models "substantially smarter than almost all humans at almost all tasks" are on track for 2026 or 2027. If that sounds dramatic, consider that the researchers building these systems believe the process has already started.
What does that mean for routine business operations? For campaign management? For financial analysis? For customer service? Those aren't hypothetical questions anymore, they're the questions every business owner should be asking themselves right now.
What You Should Actually Do About This
I'm not writing this to alarm you. I'm writing it because the single biggest advantage you can have right now is being early. Early to understand what's changed, early to start using it, early to adapt your business around it.
Start using AI for real work, not just quick questions. Sign up for Claude Pro or ChatGPT Plus. It costs about £20 a month and make sure you're using the best model available, not the default. Right now that means Claude Opus 4.6 or GPT-5.2, but the specific model changes every couple of months. The important thing is you're on the paid tier using the most capable version.
Don't just use it to ask quick questions though, that's the mistake most people make. They treat it like a slightly fancier Google and wonder what the fuss is about. Instead, push it into your actual work. If you run an eCommerce business, give it your product data and ask it to identify your most and least profitable lines. Give it your Google Ads search terms and ask it to find wasted spend. Give it your website copy and ask where your conversion messaging falls short. The first attempt might not be perfect, and that's fine. Rephrase, give it more context, try a different angle. If it even partially works today, in six months it'll do it near-perfectly. The trajectory only goes in one direction.
Think about your competitive position. The person who walks into a board meeting and says "I used AI to complete this analysis in an hour instead of three days" is going to be the most valuable person in the room. Not eventually, right now. That window of competitive advantage won't stay open forever.
Get your team experimenting. At Coffee Marketing every team member uses AI daily. Ross uses it for data analysis and campaign optimisation. Carrie uses it for client communication and account management. We didn't mandate this, it happened because the tools are genuinely useful and the results speak for themselves. The agencies and businesses that will struggle most are the ones that dismiss this as a passing fad.
Invest in your data infrastructure. AI works best when it has good data to work with. Clean product feeds, proper conversion tracking, structured website data, these aren't optional nice-to-haves anymore. They're the foundation that determines whether AI can meaningfully help your business or whether it's just guessing. If your data is messy, the outputs will be messy too. Get your house in order and the returns from AI multiply.
It's Not Just Us
What's been most eye-opening over the past few months is watching other people go through exactly the same transformation we did. Through the AI Inner Circle community and our own network, we've personally helped people with no technical background whatsoever set up Claude Code in their terminal and start building real, working projects from scratch. Not toy demos or proof-of-concepts, but actual systems that are generating revenue.
One member built a cryptocurrency trading bot that monitors markets and executes trades based on rules they defined in plain English. Another built an Amazon product research and listing system that identifies opportunities, generates optimised listings and tracks performance across their entire catalogue. Someone else created a complete task management platform for their team that replaced three different paid subscriptions. Another built a lead generation system that scrapes prospect data, enriches it with company information and feeds qualified leads directly into their CRM.
None of these people are software engineers. They're business owners, marketers and operators who learned to describe what they wanted and let AI build it. The pattern is always the same: they're sceptical at first, then they try it, then they're stunned by what comes out, and then they can't stop building. We've walked people through the initial setup, helped them structure their first projects and watched them go from "I've never opened a terminal before" to shipping working products in days. The barrier between having an idea and having a working product has essentially collapsed.
The Bigger Picture
We've focused on marketing and eCommerce here because that's our world, but the implications extend far beyond our industry. AI is now helping to build the next generation of AI. OpenAI has publicly stated that their latest model was instrumental in creating itself. Each generation of AI builds the next, which is even smarter and then the next... exponentially. The researchers call this an intelligence explosion, and the people building these systems believe the process has already started. The upside, if we navigate this well as a society, is extraordinary. Compressed timescales for medical research, scientific breakthroughs, problem-solving at scales we've never been able to attempt before. The risks are equally real though. The technology is powerful enough that getting it wrong matters enormously, and the conversation about how we handle that is one that everyone needs to be part of, not just the people building it.
What We Know
At the end of the day, here's what we know from direct experience. This isn't a fad. The technology works, it improves predictably, and the wealthiest institutions in history are committing trillions of pounds to making it better and faster. The next two to five years are going to be disorienting in ways most businesses aren't prepared for, and the people who will come out of this best are the ones who start engaging now, with curiosity rather than fear.
We've built over 40 AI-powered tools, shipped 65 platform releases and transformed how a three-person agency serves its clients. Not because we're particularly special, but because we started early and committed to learning. The tools exist right now. The question is whether you'll use them before your competitors do.
If you want to find out about how AI can improve your marketing performance, or even have us help you build your own intelligent system for your business, drop me a message or contact us using the form. No pressure, just an honest chat about what you can achieve next.
