Today at Spark AI we’ve just launched our book Shift – AI for Agencies, and I’m proud to share the thinking behind it.
My co-founder Jules Love started work in the late 1990s with Andersen Consulting, just as the internet was beginning to seriously disrupt business. Amazon was founded in 1994 and Google in 1998. Back then, we were asking ourselves whether we were better off using Netscape Navigator or Internet Explorer. Looking back, that was completely missing the point. The browser didn’t matter. What mattered was that the entire business model of how companies reached customers was being fundamentally rewritten.
Having worked with more than 50 agencies over the past 18 months through Spark AI, the company we co-founded, Jules is watching the same thing happen again. Only this time, it’s moving faster.
That’s why he wrote Shift – AI for Agencies. Not because he had all the answers, but because after 18 months on the frontline of AI adoption with agencies, he certainly knew most of the questions.
When digital arrived, agencies had years to adapt. Broadband took 10 years to roll out. The iPhone didn’t appear until 2007, and the App Store didn’t open until 2008. Agencies had two decades to experiment, learn, pivot, and rebuild their capabilities.
AI is different. More money is being poured into generative AI than any technology in human history. As a share of US GDP, it’s more than double what was spent on the Apollo moon missions in the 1960s. This level of investment means the rate of improvement we’ve seen since ChatGPT launched in late 2022 shows no sign of slowing down.
The models improve every few months. New capabilities arrive constantly. What seemed impossible last year is standard this year. And unlike the digital transition—which required new infrastructure and new devices—AI tools are immediately accessible to anyone with a laptop and a £20 monthly subscription.
Here’s what catches most agencies off guard: AI tools don’t work like the software you’re used to.
When you learned Photoshop or InDesign, you learned specific commands. Click this button, use this tool, apply this effect. The software did exactly what you told it to do, the same way, every time. Mastery meant knowing which buttons to press.
AI doesn’t work that way. The same prompt gives you different results each time. There are no buttons to press, just conversations to have. You don’t learn commands—you learn how to communicate intent, how to refine outputs, how to work iteratively with something that’s part tool, part collaborator.
This means agencies need to develop completely new capabilities. Not just “how to use ChatGPT” but how to prompt effectively, how to evaluate AI outputs, how to combine multiple AI tools into workflows, how to know when to use AI and when not to.
If AI was just about learning new tools, agencies would adapt fine. They’ve done it before. From desktop publishing to digital design to social media, the industry has consistently upskilled as new technology arrived.
But AI is different. It doesn’t just change your toolkit—it changes how creative work happens.
Think about the traditional creative process. You brief a team. They go away for days or weeks. They come back with three concepts. The client picks one. You refine it. You iterate. Eventually you have a finished brand or campaign that goes live.
Now imagine this: your strategist uses AI to analyse thousands of customer reviews in a few hours, identifying patterns that would have taken weeks to surface manually. Your creative team generates 20 concept directions in a day instead of three in a week. Your designers create 200 variations of each concept to test against different audiences. Your account team runs simulations of client presentations to anticipate objections before the meeting.
The work itself is different. It’s not about executing specific creative visions any more—it’s about designing frameworks and parameters within which AI can generate brand-appropriate content. You become the architect of creative possibility rather than the executor of individual ideas.
This is where most agencies are missing the real story. AI doesn’t just make existing tasks faster. It redistributes where time gets spent and where value lives.
Junior creatives used to spend weeks creating multiple concept directions. Now they spend hours generating hundreds of options with AI. Their value shifts from execution to curation and refinement.
Senior creatives used to spend days perfecting a single visual. Now they spend their time designing the guardrails and parameters that guide AI generation at scale.
Strategists used to spend weeks doing desk research and analysis. Now they spend that time on higher-order thinking—interpreting patterns, making connections, developing insights that AI can’t.
And here’s the critical bit: the traditional pricing model begins to break.
When creating 100 variations doesn’t cost much more than creating one, how do you charge for that work? When a task that took three days now takes three hours, why would clients pay the same rate? When work that seems—even if only from the client’s perspective—increasingly automated, why pay agency rates at all?
Any time you save through AI efficiency will eventually be passed to clients. The market will force your hand. If you try to pocket those gains as margin, someone else will undercut you.
So the agencies that are getting ahead are redefining where value comes from. They’re moving from selling deliverables to selling ongoing creative capability. From project-based pricing to outcome-based models. From “time × rate = fee” to “impact × expertise = value.”
One agency director told us: “AI hasn’t made our work cheaper, but it has made it better. Now we’re exploring more territories, testing more ideas, and then refining them more.” The agencies succeeding aren’t trying to work faster at the same things. They’re doing fundamentally different work.
Many agencies come to us asking: “What’s the best AI tool for creating social content?” Or “What’s a great prompt for writing headlines?” Or “Should we use ChatGPT or Claude?”
They’re treating AI like they treated Adobe Creative Suite, just another piece of software to master. Learn the tools, get certified, move on.
But that’s like asking in 1998 whether you should use Netscape Navigator or Internet Explorer. The browser didn’t matter. What mattered was that the entire business model of how companies reached customers was being rewritten.
The agencies getting this right are asking completely different questions:
These aren’t questions about tools or prompts. They’re questions about business transformation.
Shift – AI for Agencies captures everything we’ve learned in the last 18 months about how to transform your agency with AI. It’s built around what we call the AI Maturity Model—a four-stage framework that walks you through how to move beyond scattered experimentation to structured adoption.
The book covers how to develop an AI-forward culture, deploy AI-enabled workflows, put the right guardrails in place for responsible use, navigate client conversations that protect your value, and rebuild your value proposition using AI to amplify what makes your agency special, not dilute it.
Jules studied Applied Generative AI at MIT before we co-founded Spark AI to help agencies navigate this future. We were supported by Innovate UK to build our AI Accelerator programme, and Shift contains everything we’ve learned from the frontline of AI adoption. Many of the concepts in this book Jules teaches as part of the Advanced Diploma for AI in Business at Oxford University’s Saïd Business School.
The agencies that will thrive aren’t the ones with the biggest budgets or the most advanced technical resources—they’re the ones that understand they’re not just adopting new technology, they’re rebuilding what an agency is, what it does, and how it creates value.
Order ‘Shift – AI for Agencies’ to transform your business for the AI era. To celebrate it’s launch on 4th November, BIMA members will be able to buy the Kindle edition for 99p for the first 48 hours. Order it on Amazon here: https://www.amazon.co.uk/dp/1781339511/
About Spark AI:
Spark is an AI training, coaching and consultancy working with agencies and brands. We help leaders upskill their teams, build AI workflows and reconfigure their businesses. We have been supported by Innovate UK and teach our AI for Leaders programme at Oxford University’s Said Business School.