👋 Greetings earthlings,

Emma here, co-founder of Spark AI. This is our last big edition of the year, and what a year it’s been. Before we dive into 2026 predictions and our usual practical agency-specific tips, we’ve got something for you:

📅 Live webinar and Q&A with Jules tomorrow – my co-founder (and husband) and author of industry #1 bestseller Shift – AI for Agencies. Register here

But first, a moment to pause and say thank you. I'm so proud of what Spark Intelligence has become. It's something I genuinely love creating, and that's down to you. The open rates, the replies, the thoughtful feedback, the "I forwarded this to my whole team" messages. You've turned a fortnightly email into a real conversation. Thank you for being part of it.

At Spark AI this year has been anything but dull: hugely satisfying, mind-bendingly challenging, and completely absorbing. Running a business in a space that reinvents itself every fortnight is hard, doing it while staying true to your values is harder! But having people around you who genuinely care about the our clients makes it not just possible, but a privilege and a genuine mission. Thank you, Team Spark. I'm so proud of what we've built this year.

Now back to it, kicking off with a celebration of how the industry as a whole has moved so far forward with AI this year. Let's get into it.

• 1. A year in the life of AI in Agencies

• 2. Just build an app for that!

• 3. Tool drops: Chat GPT 5.2, Abobe and Disney

• 4. Join us for a LinkedIn Live Q&A with Jules

1. A year in the life of AI in Agencies - what we actually went through with AI this year

Looking back at 2025, I’ve been thinking about the emotional journey most agency teams have been on with AI over the last year. It’s been quite a ride, there’s been really hard graft and truly inspirational entrepreneurial thinking. So I wanted to take a moment to massively celebrate the progress of the industry, leading the fore with AI adoption, and chart what I’ve seen from the unique perspective of looking across many businesses:

🗓️ January: “I know AI matters, but where do I even start? It feels like every else has figured it out already”

Most people I spoke to began the year stretched thin. AI felt important, but also overwhelming. They wanted space to get their heads around it without feeling behind.

Many people are at this stage right now – we are all moving at different speeds depending on what’s right for our businesses.

☀️ Summer: “We’re trying loads of tools… but it’s chaos. Everyone has their own way of doing things and I can’t tell what’s actually working”

By the middle of the year, everything opened up. People were experimenting with tools, sharing prompts, testing ideas, and generally giving things a go. They subscribed to an LMM for the agency, then let their teams just go and experiment… And then the scatter arrived.

This stage is progress – but it’s also the moment where:

• quality fluctuates - do our team know how to use it?

• risk creeps in - are the tools we’re using safe?

• delivery begins to slow down - where is AI actually helping us?

• and leadership starts to worry about the lack of alignment - how do we get it working for us?

No one had done anything wrong – this is just what happens when interest outpaces structure. Quality varies, everyone has their own approach, and before long it becomes hard to tell what’s producing value and what’s producing noise.

This is exactly where our AI Fundamentals workshops come in – designed for teams who are figuring out where to start, and how to build shared language and confidence. We’ve run 42 this year – let us know if you'd like to book one for 2026.

Mark Horley, Chief Creative Officer and Co-founder of award-winning employer branding agency Tonic was quite clear about what thought after his workshop:

🎄 December: “We’ve proved the case for AI - everyone’s using it - now we need to make our AI use consistent so it adds value to the business.”

As AI started showing up inside our everyday tools, things shifted quickly. Suddenly it wasn’t an optional experiment – it was part of the workflow. The questions we were getting changed:

• How do we make this consistent?

• How do we keep quality high?

• How do we avoid reinventing the wheel every day?

This is where our AI Accelerator comes in – designed for agencies ready to move from “we’re experimenting” to “this works across the whole business.”

Over 3 months, we help teams:

• Upskill confidently – and turn hit-and-miss results into real advantage

• Build repeatable AI workflows – so everyone’s pulling in the same direction

• Align sceptics and believers – and create shared language and goals

• Connect AI to commercial outcomes – not just creative output

• Tame the chaos – and replace scattered experiments with a structured plan

Agencies like BWP, Omne, Analogue and mark-making* have used the Accelerator to move from enthusiasm to embedded capability this year.

TLDR

Across the year, most teams moved through three very human stages:

Trying to understand it → Trying things out → Trying to make it consistent

But the truth is most organisations are still somewhere in the first two. The later stages don’t just “happen” because people have access to tools. They happen when leadership treats AI as business transformation and allocate resource, time, upskilling support, and a clear way of working. That’s what 2026 will demand.

Looking ahead to 2026

This year, the majority agencies moved from curiosity to experimentation with Ai. They bought licenses, ran workshops, got a few enthusiasts prompting well. That was progress. But it wasn't transformation.

A much smaller group, fewer than 20% according to our most recent research, did something different. They didn't just try AI. They rewired how they work. Full workflows redesigned around AI. New skills hired and developed. Some productised their IP and built their own tools. AI is just part of how they operate.

Here's what I think happens next: that gap widens.

This year the agencies with entrepreneurial leadership and agile business models moved first. They'll keep pulling ahead and many more will join them, making the leap from scattered experimentation to genuine operational change. But plenty won't. They'll stay busy with pockets of enthusiasm and no consistency, hoping AI implementation will happen organically.

The difference between these two groups won't be tools or talent. It will be structure they put around AI adoption. The agencies that build real capability across their teams, that set clear expectations, that create workflows people actually follow, will see it in their margins, their quality and their ability to deliver predictably.

I’m expecting to see more of a quiet divergence than a dramatic disruption in 2026, but that’s perhaps more dangerous. So my plea to you in 2026 is please set your brilliant creative and curious minds to how your business can leverage AI to create value. If you need help, this is exactly the challenge we founded Spark to support you with.

2. Just build an app for that!

Last week I turned a messy notebook sketch into a fully interactive financial model in under 60 seconds. No spreadsheet.

This is a shift I want more people to understand: AI isn't just a writing assistant anymore. The emerging Generative UI capability of our AI tools means rather than just answering your question, it creates a working tool to solve your problem. You can turn any back-of-the-napkin idea into a clickable prototype before your coffee goes cold.

Last week I was testing this with live business challenges using Gemini Canvas mode (Claude does this brilliantly too). Let me show you exactly what I mean.

What I built: interactive project team cost modeller

My old workflow for modelling project resource costs went something like this: sketch a structure on paper, open Excel, build formulas, manually enter data, realise you made an error in row 47, start again. It worked, but it was slow. The new workflow: Photo → Prompt → Interactive Model. Here's how it worked when I tested it this week.

From sketch to financial model

First I sketched out a potential team structure for a new project in my notebook (for the newsletter I’ve created a fictional "Project Alpha"). It’s messy, it mixes full-time staff with freelancers, and you aren't sure if the budget holds up.

My workflow:

1. Take a photo: Take a picture of your notebook sketch of your team chart.

2. Upload to Gemini: Drop the image into the chat.

3. The prompt: I simply asked "Write this project alpha team structure up and create an interactive app I can model the cost of this mixed freelance and employed team and analyse runway."

What happens next:

• Gemini decodes the image: It reads your handwriting (even mine), identifying that "Theo" leads Commercial and "Alice" is on a day rate.

• It builds the logic: It understands that "Runway" = "Starting Cash / Monthly Burn" and writes the code to calculate this.

• It deploys the tool: Instead of text, it gives you a fully functional Cost Modeller App right in the chat window. From here you can download this to desktop or create a live web app for your team.

The result: 

You get an interactive dashboard where you can:

• Toggle team members between Employee (Salary + Benefits) and Freelancer (Day Rate).

• Adjust days-per-month for the freelancers to see how part-time vs. full-time affects the bottom line.

• Input your starting budget to instantly visualise your runway (e.g., "If I hire Alice full-time, we run out of cash in Month 8. If she is 3 days/week, we last until Month 14").

It turns a static thought into a dynamic scenario plan in under 60 seconds. Give it a go – and let me know what you think and what other applications it leads you down a rabbit hole of building… I’ve got a feeling my Christmas break might be busy building!

3. Tool drops: This week belongs to ChatGPT

A couple of weeks ago all eyes were on Google Gemini’s big update and Sam Altman calling a code red at ChatGPT in response. Right on the back on that ChatGPT 5.2 launched and they signed a deal with Adobe and Disney.

I haven’t had time to stress-test GPT-5.2 properly yet, so rather than guess, I’m leaning on two voices I trust to cut through early hype: Ethan Mollick and Allie K Miller. Their reactions help explain why this release matters, especially if you’re already a ChatGPT user.

GPT-5.2: ChatGPT grows up

GPT-5.2 looks like a meaningful upgrade for people who use ChatGPT for real work, not just quick answers.

Mollick points to GDPval, one of the most economically relevant benchmarks of AI we have. On tasks that take humans 4–8 hours, GPT-5.2 beats human experts 71% of the time. Previous models struggled to cross 50%. Suggesting AI is now competitive on the kinds of knowledge work agencies and marketing teams actually charge for.

Allie K Miller’s early testing backs this up. She describes noticeably stronger reasoning, deeper problem exploration, and a model that sticks with complexity rather than skating over it. The trade-off? Outputs can be long and heavily structured. Great for analysis and specs. Less great when you just want a short answer to paste into an email. This feels less like “AI as a chatty assistant” and more like AI as a serious analyst doing real tasks like planning, analysis and problem-solving. Read more.

Disney: from fighting AI to licensing it

At the same time, Disney quietly signalled where the industry is heading. Its reported $1bn deal with OpenAI brings hundreds of Disney, Marvel, Pixar, and Star Wars characters into video model Sora, while keeping tight controls around training and actor likenesses. A major shift from litigation to licensing.

This points to a future where brand characters and assets are expected to live inside AI tools, with clear rules and guardrails. Read more.

Adobe: strategic move for new customers

Adobe have quietly put Photoshop and Acrobat inside ChatGPT, giving it new consumer reach at massive scale. This means ChatGPT becomes the interface people start from and Adobe remains the infrastructure brands rely on to actually get work done safely. Read more.

4. Join us for a LinkedIn Live Q&A with Jules

Tomorrow we are hosting a LinkedIn Live where Jules will unpack what’s behind his bestselling book Shift – AI for Agencies*:

• The patterns he's seen in agencies that are genuinely moving forward with AI

• Why most agency AI initiatives stall (and what to do instead)

• The questions agency leaders should be asking right now

• Live Q&A on anything from the book or your own AI challenges

Whether you've read the book or not, this is a chance to dig into the strategic side of AI adoption with someone who's been in the trenches with dozens of agencies this year.

📅 16th December 🕐 11am 📍 LinkedIn Live

👉 Register here

Can't make it live? Register anyway and we'll send you the recording.

*Shift – AI for Agencies is available on Amazon at Waterstones or Hatchards (and all other good booksellers)

Until next time (actually, next year!)

That’s all for today and for 2025. I hope you’ve enjoyed Spark Intelligence this year, it’s been a real labour of love for me. Mostly written on Sundays, interrupted by dog walks where I get yet another idea or two to add in. I’d love to know if you’d like more of the same or different next year – just hit reply or use the poll below.

Heartfelt thanks for you for all your support this year. And all our very best for 2026 from all of us at Spark AI. See you then!

Emma 

Co-founder and CEO of Spark AI

About Spark AI

Spark AI empowers creative and brand leaders turn AI curiosity into confidence through structured training and business transformation.

We have worked with 60+ agencies running AI Fundamentals workshops and AI Accelerator programmes based on our # 1 bestselling book Shift – AI for Agencies.

Trusted by Oxford University Saïd Business School and backed by Innovate UK.

👉 Find out more 

Not a subscriber yet?