It’s a tough time to be a marketer. The overwhelming online narrative is that there are many ways you’ll lose your job and become obsolete. Your job disappears if you’re not AI-native; if you’re not a systems builder. AI means you don’t need a CMO or even a marketing leader.
The above narrative is everywhere. If you’re a marketer trying to build a career, it’s intense.
You’ll hear time and again - you need to become a systems builder. I’ve even said this myself.
But what does it even mean?
For people from a non-engineer or technical background, it can sound like AI jargon.
I want to demystify it and help people create their v1 marketing system.
My relatively straightforward explanation of a systems builder is: “A systems builder can look at repeatable work and turn it into a system where AI handles parts of the thinking and execution.”
That’s it.
For a marketer, that could be competitor monitoring, campaign creation, customer research, reporting, content creation, preparing for 1:1s, or figuring out what you should work on today.
You don’t need to become an engineer. You need to get good at spotting repeatable work and asking: can AI make this faster, better, or cheaper?
Let’s take competitor research as a simple example: a very common marketing task, and one you can definitely build a ‘system’ around.
The Non-System Approach
All marketing teams have to do competitor research periodically. Check what competitors are doing, open a bunch of tabs, read product pages, case studies, and recent launches on social media, look at their ad creative, dump some observations into a document, do an exec summary, and send it to a bunch of people.
The following month, you have to repeat the same process.
The Systems Builder Approach
A systems builder looks at the process differently. Could I build an intelligence layer that already knows who our competitors are, our ICP, and our positioning? Could I build a skill that continually updates that intelligence with new launches, messaging, case studies, pricing, and advertising? Could I then build skills on top of that intelligence to help with positioning, campaigns, sales enablement, and product strategy?
Now you are not repeatedly doing competitor research. You’ve built a competitor intelligence system that compounds over time.
As a systems builder, you’re constantly asking, how do I make this work reusable?
If you’re a marketer and want to build your very first ‘AI system’, I have a starter one for you.
The Starter AI Marketing System
I’ve built a V1 Starter AI Marketing System that you can copy and customise.
Versions of many of these things live inside my own system. Mine is a little more complex and, at times, wildly over-engineered :)
This is where I would start.
The system has 5 layers:
Intelligence: the context everything else reads from
More of You: skills that scale how you work
Better Thinking: skills that challenge your judgement
Product Story: skills that help you understand customers, competitors and differentiation
Run My Day: skills that help you focus on the work that matters
I’m also giving you the prompt to build the whole thing.
→ [COPY THE STARTER AI MARKETING SYSTEM]
Let’s go layer by layer.
Layer 1: Intelligence
This is the true game changer and what separates any system you build from a series of prompts. If you want AI to produce consistently good work for you, it needs to understand the world that work exists inside of.
Take the time to build your context/intelligence.
I’d start with 6 core intelligence files you can build
1. Goals
What are you trying to accomplish this quarter? What metrics matter? What are the priorities? Where are you behind? What explicitly does not matter right now?
This becomes incredibly useful because every skill can anchor its recommendations to what you’re actually trying to achieve.
2. ICP
Who is your ideal customer? What problems are they trying to solve? What language do they use? Why do they buy? Why don’t they buy?
Your ICP file should continually improve as you learn more from sales calls, support tickets, interviews and product usage. This is what sets it apart from traditional marketing teams with ICPs stuck in PowerPoints that no one reads.
3. Competitors
Who are the competitors you care about? How are they positioned? What have they launched? What are their latest case studies? How has their paid messaging changed?
Again, this should not be a static document you create once and forget.
The system should keep it current.
4. Positioning
Why should someone choose you?
What do you stand for?
Where are you genuinely different?
What claims can you credibly own?
This becomes the underlying context for a lot of the marketing work you do. In today’s noisy world, your product story is everything.
5. Team
If you’re a marketing leader, who reports to you? What are they responsible for? What are their goals? What are their current challenges?
This becomes extremely useful for management skills we’ll build later.
6. Executives
Who are the people you need to influence? What do they care about? How do they make decisions? What questions do they repeatedly ask?
For some people, this may simply be your manager.
The point is to capture context about the people you need to communicate with.
What this replaces: repeatedly explaining your business to AI, repeatedly uploading docs that are likely outdated for the AI to pull context from.
Once these files exist, every skill you build on top becomes better.
There is one important detail.
I don’t want every skill reading every intelligence file.
That becomes clunky quickly. The model has to hold too much irrelevant context, uses more tokens, and can get confused.
Instead, skills should only pull the context they need.
A competitor analysis skill might read your competitors, ICP, and positioning.
A 1:1 skill might read your goals and team context.
Think of it as routing intelligence through your system.
→ [GET THE INTELLIGENCE LAYER + BUILD PROMPT]
Add an AI coach
The next thing I put inside the foundation of the system is an AI Coach.
In ChatGPT, make sure Memory and Reference Chat History are enabled. In Claude, use Memory or keep this inside a Project where Claude can build context about how you work.
Then give it a recurring job:
“Look across how I’ve been using you and coach me on how I could use AI better.”
A couple of good questions to ask are
What are 3 repetitive tasks I keep asking AI to do that should become skills?
Which files in my intelligence layer probably need updating?
What feature am I not using that fits how I actually work?
Why does this work so well?
Because the model already has a compressed representation of how you work. It has seen the kinds of problems you bring it, where you repeatedly get stuck, what you keep explaining again and what tasks keep showing up.
It does not have a perfect replay of every conversation. Both ChatGPT and Claude selectively retain useful context rather than carrying every historical detail into every chat.
For this use case, that’s fine.
You don’t need an audit log. You need the patterns.
If I keep asking AI to turn messy notes into executive updates, that probably wants to become a skill.
If I repeatedly explain who my ICP is, my intelligence layer is missing something.
If my customer research has started contradicting icp.md, that file probably needs an update.
And if Claude or ChatGPT launches a capability that maps directly to work I already do, my coach can point that out.
Record the way you work
You also need a repeatable way to build skills to help you do your work. The easiest way of doing this is to record yourself doing the work.
Claude now has functionality that allows you to screen record yourself performing a task and talking through what you are doing, then use that process to create a skill.
That means you can simply now just record yourself work and ask Claude to build that into a skill.
Layer 2: More of You
Layer 2 is all about using AI to extend your own capabilities to have more impact. I’d start with 3 skills.
1. People Brief
I want my 1:1s to be a good use of time.
So the People Brief reads from my goals and team context and prepares me before each conversation.
It can tell me what we agreed last time, what follow-ups I personally owe them, how they are tracking against goals, what blockers exist and what wins I should recognise.
Because I’m continually writing context to my intelligence layer, that intelligence layer tells me in advance of these meetings what are the most important things to spend time on.
2. Voice of You
The voice of you skill may seem a little dystopian, but it allows people to interact with a synthetic version of you to get thoughts, opinions, pushback, and clarity before actually coming to you. The purpose of it is to make sure that when they actually do interact with you, they have fully formed questions and can focus on the most important aspects of the problem, opportunity or work.
Because over time your context layer is gathering intelligence about how you make decisions and your work, it can actually get pretty accurate at giving someone a first version of your perspective on their work, their problem, their opportunity, and how they can improve things
3. The Storyteller
The most important part of a marketing leader’s job, but really any marketer, is to continually sell the value of marketing internally. Marketing is one of the more difficult jobs because it’s not always easy to see the return on investment.
You have to be your own best storyteller, and this skill helps you translate your outcomes into stories that you can tell internally so every other team understands the value that you are bringing to that role.
Layer 3: Better Thinking
Some of the best uses of AI aren’t to continually execute and create things. They’re actually to make you a better thinker. The skills in this layer are to help me be more impactful in my work and help actually better fine-tune my critical thinking and how I solve problems, or how I execute as a marketer, a growth leader, and go after opportunities.
1. Exec Panel
Just like you had a synthetic version of yourself for your team, colleagues, people who work with you, this is a synthetic version of your execs, your manager, or the people you have to manage up to. Managing up is one of the most important things you have to do as a marketer, whether it’s to a marketing manager, a marketing leader, or you are managing up to the C-suite execs.
The cool thing about AI is that, because you can capture all of the information the execs are creating internally, you can create a little executive panel: a synthetic version of your execs. You can pitch them things, show them things, and present to them to get feedback before you ever have to do the real thing. Pretty cool use of AI
2. Challenge Me
AI is a great tool to run different frameworks to challenge your thinking. Mine has:
Inversion: What would guarantee this fails?
Pre-mortem: It’s 6 months from now and this failed. Write the post-mortem.
Assumptions: Which assumptions does this strategy rely on that we haven’t validated?
Second-order effects: If this works, what happens next that we aren’t accounting for?
I’d start with one framework you like and just use that one.
Layer 4: Product Story
I deliberately put Product Story into the starter version of this system. If you are a demand marketer, performance marketer or SEO, you may swap this layer for something closer to your own work.
But I think product story is one of the most important things marketing teams need to have right now. Everything is so noisy.
Your product story needs to speak with your customers in a way that sounds different from everything else.
1. Voice of the Customer
The voice of the customer can be built from internal and external data. Yes, we are going back to more synthetic versions of things. The problem with the traditional ICP that most companies use to do some sort of fit and intent model is that it gets outdated pretty rapidly. It lives in a PowerPoint, it lives in a Google Doc, and it lives in some sort of place that no one ever really pays attention to. It gets outdated and isn’t usable.
The cool thing about doing a voice of the customer is that you can actually build a skill that continually updates it from all of the things it’s learning about your customer. And of course, you’re continually recording new learnings about your customers to the intelligence layer.
This is a true synthetic version of your customer that you can actually speak to and pitch new product ideas to, product positioning things like that, and actually get real feedback from your customers.
Now, it doesn’t replace speaking to customers. I cannot emphasise that enough. Speaking to customers is the number one job of any great marketer, but it is certainly a great way to gather research pretty rapidly
2. Voice of the Competitor
The voice of your competitor is pretty similar. This is the research skill that we talked about right at the start of this post. It’s able to go and scrape competitors’ websites, look at their positioning, look at the product updates. It can go ingest paid advertising messaging, ingest the latest case studies, and look at what they’re posting on social. It can actually give you the most important things that you need to know about:
How are they positioning against us?
Are there customer segments moving at all?
Are they changing the value prop?
This gives you a really great perspective on the current competitive landscape that you’re in.
3. Are We Really Different?
Now you can have a skill that reads your positioning, ICP and competitor intelligence and grades whether you actually sound different.
Give you a core from 1 to 10. Some of the fun questions to ask are:
Show me which claims every competitor can also make.
Show me where our language has drifted towards the category norm. e.g. it’s just all the same
Show me where we have a credible position nobody else owns e.g. where can we sound different.
You need a system that helps you fight back against the tyranny of sameness.
Layer 5: Run My Day
The last layer is how I use AI to run my day.
1. Priority List
This skill reads my goals, team context, executive needs and the relevant information coming through channels like email and Slack.
Then it tells me what I should work on.
Mine’s priorities are someone blocked waiting for a decision, a KPI that has moved off target, an important follow-up I owe or something an executive needs from me.
The key is prioritisation. You don’t want AI to just give you everything, you want to teach it what matters to you.
2. What Did I Miss?
This one looks for things I owe people - Emails, Slack messages, Meeting follow-ups, Google Doc comments etc.
Then it groups them by importance and how long people have been waiting and gives me enough context to respond.
3. Blocker Watcher
As a leader, one of the highest-value things I can do is unblock people.
So I want AI looking for signals like people waiting on other teams, missing decisions, missing budget, missing headcount, missing data and repeated slip dates.
Then tell me: Who is blocked? How long have they been blocked? What is causing the problem? What can I personally do to help?
I’ve found this has been hugely impactful.
4. Custom Dashboards
Finally, use AI to change how information gets presented to you. Custom dashboards is a tier 1 use case for AI.
Aggregate the data you care about and build interfaces around the questions you actually need answered. Claude and ChatGPT are perfect for building live artifacts around custom data.
Copy the whole system
This is a good starting system for you to build on top of. You can simple go here, copy the prompt and start asking ChatGPT or Claude or personalise to your needs.
So, when someone online tells you:
Systems builders are going to survive AI.
You should now have a much clearer idea of what that means and a system of your own to amaze them with :)
Until Next Time,
Happy AI’fying
Kieran