In this ChatGPT Super-User Guide, we’ll look at how to use ChatGPT for business in a way that actually builds momentum…
The Shift Most People Can Feel (But Can’t Quite Explain)
If you’ve started using ChatGPT, you’ve probably had that slightly confusing experience where part of you is impressed, and part of you is wondering what you’re missing. Because it is helpful. You can ask it to write something, organise something, explain something – and it responds quickly, sometimes surprisingly well. But then you come back the next day, open a new chat, or try to build something more substantial with it, and it all feels a bit disconnected again. Like nothing is really building. Most people accept that as part of the experience. They assume that’s just how AI works. But it isn’t.
What’s happening isn’t that AI lacks intelligence. It’s that it lacks continuity – and most people never give it enough context to create any. So they dip in and out. Ask a question. Get an answer. Move on. Which makes it useful – but not powerful.
The people getting real leverage from AI aren’t asking better questions. They’re working differently. They’ve stopped treating AI like a place to go for answers, and started treating it like a place where their thinking lives. That shift sounds small. It isn’t. Because once your thinking has somewhere to sit, it can actually build.
AI doesn’t struggle because it lacks intelligence. It struggles because it lacks context—and most people never give it enough.
From Tool to System
Why Most AI Use Plateaus
In the early stages, AI feels productive because it gives quick wins. You can draft a landing page, generate ideas, or get unstuck on something small. But nothing carries forward. Each interaction resets. For experts – coaches, consultants, educators – this is where the friction begins. You’re used to building ideas over time. You revisit them, refine them, connect them into something coherent. But with AI, the moment you close a chat, that thread disappears. So instead of momentum, you get repetition. This is often where founders hit what I call the founder bottleneck – where everything still depends on you holding the thinking together. You explain the same thing again. You rethink decisions you’ve already made. You rebuild context that should already exist. That’s where most people plateau. Not because the tool stops working – but because their way of working never evolves. This is where most people struggle when learning how to use ChatGPT for business.
What Changes When You Build a System
The shift is not about learning more features. It’s about changing how you approach the work. Instead of asking: “How do I get a better answer?” You start asking: “How do I stop restarting my thinking?” That’s the real problem.
Once you solve that, everything changes. You stop chasing perfect outputs in a single moment and start shaping ideas over time. Your work becomes continuous rather than fragmented. And AI starts to feel less like something you use occasionally, and more like something that is quietly supporting how you think every day.
The real limitation isn’t the tool. It’s restarting your thinking every time you use it.
How ChatGPT Actually Works
Keeping It Simple So It Actually Works
A lot of confusion around AI comes from language. Projects, chats, (custom) GPTs, agents – it all starts to blur together. But the structure itself is simple.
- A Project is what you are working on. It’s the container for your thinking.
- A Chat is where the work happens. It’s the conversation where ideas are explored and developed.
- A Custom GPT (sometimes just refered to as GPT) is how you are supported. It represents the type of thinking you bring into that conversation.
If you’re building a premium programme, for example, that programme is your Project. Inside it, you might have ongoing conversations around strategy, messaging, and delivery. And at different points, you may want different types of thinking – strategic, creative, analytical – to support you. Keeping these layers clear doesn’t add complexity. It removes it. Because now your thinking has somewhere to go – and somewhere to return to.
Projects: Where Thinking Lives
A Project is not a task. It’s something that continues.
If you are building an online course, refining your offer, or shaping your positioning, you are not completing something in one sitting. You are making a series of decisions over time. Each one builds on the last. When you treat that as a single, ongoing space, your work starts to feel coherent. You are no longer jumping between disconnected fragments. You are developing something.
A common mistake is trying to organise everything too neatly. One Project for content, one for marketing, one for offers. It feels structured, but it breaks continuity. You end up splitting your thinking across multiple places, which means nothing fully develops.
A simpler approach is to create a small number of meaningful containers that reflect the real parts of your business, and allow your thinking to build within them. That’s where momentum comes from – not from doing more, but from not losing what you’ve already done. This is also where having a clear structure becomes invaluable. It’s exactly the kind of work we map out inside a Blueprint Intensive – so your thinking isn’t just captured, but translated into something you can actually build.
Momentum doesn’t come from doing more. It comes from not losing what you’ve already thought.
How Memory Actually Works
Most confusion with AI doesn’t come from the tool itself, it comes from not understanding how memory is handled. Once you see the three layers clearly, everything becomes easier to control and far more predictable.
Chat Memory (Temporary)
This is the most immediate layer. It lives inside a single conversation and holds everything you’ve said and received in that thread. It’s what allows the AI to follow along, refer back, and stay relevant as the discussion evolves. But it doesn’t exist outside that chat. When you start a new conversation, this context is gone. This is why a great result in one chat doesn’t automatically carry over to the next.
Project Context (Semi-Persistent)
This sits one level higher. Within a Project, you can add instructions, upload files, and build multiple chats that all relate to the same piece of work. This creates continuity. You’re no longer starting from zero each time, you’re returning to an environment that already understands your goals, your language, and your structure. However, this context is contained within that Project. It doesn’t automatically transfer elsewhere.
Personal Memory (Persistent Across Everything)
This is the deepest layer. It sits outside any single Project and shapes how the AI responds to you over time. This is where your preferences, patterns, and ways of working begin to accumulate. You can think of this as your “Personal DNA” a baseline that influences future conversations without needing to be restated each time. You can store your “Personal DNA” by giving the AI context about how you think and work, and asking it to remember it so that over time, it responds in a way that feels more aligned to you.
Remember your “Personal DNA” works best when it only holds stable information, not everything about your work. This means capturing what doesn’t change often – who you serve, how you think, your working style, your strengths, and how you prefer responses, so the AI can stay aligned with you across conversations. What should not go here is anything that changes regularly, such as your short term goals, current business priorities, specific project direction, or short-term strategy. These belong in your Projects or individual chats, where they can evolve without creating confusion. A simple way to hold this is: if it changes often, don’t store it globally; if it defines how you think, it belongs in your Personal DNA.
To update persistent memory you can click your profile picture or initials in the bottom-left corner, then click Settings, go to Personalisation and open Memory. There you can review saved memories and delete individual memories.
Chats: Separating Your Thinking
A simple way to get better results from ChatGPT is to treat chats as separate spaces for different types of thinking. Instead of opening a new chat for every task, keep fewer chats and return to them.
Think in terms of intent, not tasks. For example:
- Strategy Chat → direction, decisions, trade-offs
- Execution Chat → next steps and planning
- Analysis Chat → reviewing results and insights
- Production Chat → creating emails, pages, documents
Each chat holds its own context. When you come back, you pick up where you left off instead of starting again.
It’s important to also note that context doesn’t automatically carry between chats, even when they are in the same Project. Each chat remembers its own history. That’s why something you discussed in one chat may not appear in another. The fix is simple: either return to the same chat, or start a new one with a short context summary.
However, in some cases, the Ai might use important details you’ve shared previously, or preferences or patterns from earlier conversations – but it’s selective. Sometimes when you open a “new” chat, if it’s part of the same session or flow, there can be carryover context too. At times this might feel like Ai knows everythin – and at other times that it has frustratingly forgotten. The issue is that this can be unpredictable and it’s best to realise one chat cannot see the content within another.
Also, if things start to feel messy within a chat, it’s time to reset. Signs to look for are that the answers become generic or repetitive, the conversation drifts away from your goal, you’ve mixed too many topics in one place etc. Starting fresh is how you keep clarity, and the best way is to open a new chat and begin with a short context block:
What I’m working on
What I want to achieve
What matters most right now
Any key constraints
Keep it brief. You’re just grounding the conversation so the AI doesn’t have to guess.
Used this way, chats become a simple system. You keep your thinking organised, reset when needed, and avoid the slow drift that comes from trying to do everything in one place.
Giving AI Direction
Why Outputs Feel Generic
One of the most common frustrations with AI is that the output feels fine – but not quite right. It reads well. It sounds coherent. But it doesn’t feel like it belongs to your work. It feels like something that could apply to anyone. That’s usually because it has been created without enough context.
AI doesn’t struggle because it lacks intelligence. It struggles because it lacks perspective. And when it doesn’t have enough perspective, it fills the gaps with what is broadly true. That’s why so much AI output feels generic. It isn’t wrong. It’s just not specific enough to be useful. You can recognise this when something feels polished but slightly detached. It doesn’t reflect your audience, your stage, or the decisions you’ve already made.
Generic output isn’t a quality problem. It’s a context problem.
What Better Direction Actually Looks Like
A small shift fixes this. Before you begin, take a moment to anchor what’s actually happening. Not in a formal way – just in a real one. For example: You’re not “writing a landing page.” You’re refining a premium offer for experienced professionals who are cautious about online business. You’ve already tested the idea. You’re now trying to improve clarity and trust without sounding overly sales-driven. That level of context takes seconds to express. But it changes everything. Because now the AI isn’t guessing. It’s responding to something real.
One of the most useful (and often overlooked) parts of working inside a Project in ChatGPT is the ability to add persistent instructions and reference material. Under the Project’s settings, you can define how you want the AI to behave and upload files that it can draw on as context (e.g. your communication style, your brand ethos etc). In practice, this means you’re no longer re-explaining your business, your audience, or your preferences every time you start a conversation – the Project holds that for you.
Simple text files work best here. Things like your positioning, audience notes, offer structure, brand voice, or past strategy decisions can all be stored and referenced. Over time, this creates a stable foundation for your work. Instead of starting from a blank page, you’re returning to a context that already understands what you’re building. It’s a small shift, but it’s one of the key ways Projects move from being a place you chat, to a place where your thinking actually lives and evolves.
Why Prompts Fall Short
There is a lot of focus on prompt writing and prompt “engineering”. Templates, formulas, and “perfect prompts” are everywhere. But in practice, prompting is far simpler than it is often made out to be. A prompt is just a way of giving direction. And the quality of that direction depends far more on your clarity than your wording.
When prompts don’t work, it’s rarely because they’re badly written. It’s because something important is missing. The AI doesn’t know enough about what you’re trying to do, where you are in the process, or what matters most. Once that clarity is in place, prompting becomes lighter. You stop trying to force a perfect answer out of a single instruction. You start building within a context that already understands your work. That’s when it begins to feel natural.
One of the simplest ways to improve your outputs is to treat each prompt like a short briefing. Rather than relying on the AI to fill in gaps, you provide the context it needs up front. This is what is meant by content injection. You clarify who this is for (the relationship and audience), what’s happened before (the backstory or current situation), what you want to happen next (the goal), and the tone required (based on the moment, not just your brand voice). It doesn’t need to be long or formal, but it needs to be clear. When you do this, the AI stops guessing and starts working with you – producing outputs that feel grounded, relevant, and far closer to what you would have created yourself.
Better prompts don’t start with better wording. They start with clearer thinking.
Choosing How Deep the AI Thinks
If you’re using ChatGPT, you may have noticed options like Instant and Thinking, along with settings that adjust how much effort the AI puts into a response. This can feel like a small detail, but it has a real impact on the quality of what you get back.
At a simple level, this setting controls how deeply the AI works through your request.
Instant is faster and more direct. It’s useful for quick tasks, simple questions, or when you already know what you’re asking for.
Thinking takes more time, but produces more considered responses. It’s better for strategy, problem-solving, and anything where the answer isn’t obvious.
You may also see options to increase the “thinking effort.” This allows the AI to spend more time working through the problem, which can lead to more structured and thoughtful outputs, especially for complex or ambiguous tasks.
What matters here isn’t choosing the “best” option, but choosing the right one for the situation.
If you’re asking for a quick rewrite or a simple answer, speed is often enough. But if you’re working through something that requires judgement – like refining an offer, thinking through a decision, or structuring a new idea – it’s worth slowing things down and letting the AI think more deeply.
This ties directly back to how you use AI overall. Better results don’t come from one perfect prompt, they come from matching the level of thinking to the task in front of you.
Custom GPTs: Consistent Thinking
If you keep repeating how you want the AI to think, you don’t need better prompts - you need a better starting point.
When They Actually Become Useful
Most people don’t need Custom GPTs at the beginning. They become useful once you notice repetition in your thinking. For example, if you regularly find yourself saying:
- “Make this clearer but keep the tone grounded”
- “Challenge this idea rather than agreeing with it”
- “Help me think strategically, not tactically”
Then you are already defining a way of working.
A Custom GPT simply holds that for you. So instead of restating it every time, you start from that position automatically.
Custom GPTs are not like team members. They don’t take ownership of tasks or move work forward over time. They don’t “go away and do the work” for you. They are closer to bringing in a specialist perspective at the moment you need it. For example, imagine you’re working with a copywriter. You sit down together, explain what you’re trying to achieve, and they help shape the messaging, refine the wording, and improve clarity. But once that conversation ends, they’re not continuing the work in the background unless you actively engage them again.
A Custom GPT works in the same way. It brings a specific way of thinking into the conversation- whether that’s strategic, analytical, or creative – but it stays within that moment. It influences how something is done, not what happens next. So rather than thinking of Custom GPTs as people doing the work, it’s more helpful and accurate to see them as the expertise you bring into the room when you need it – the “hat” you might need to wear for that task like a highly skilled copywriter or analyst you might consult.
Explore Custom GPTs: Where You Start Building Your Own
If you’re using ChatGPT, you’ll notice an “Explore GPTs” tab. When you click it, you’re taken to a space where you can browse existing Custom GPTs and create your own. Creating one is straightforward: you define what you want it to do, describe how it should behave, and give it any context or examples it needs. What you’re really doing is shaping a repeatable way of working – capturing instructions and patterns so the AI responds consistently. From there, you may even begin to push it further toward agent-like behaviour by asking it to break tasks into steps, carry work forward, and refine outputs over time.
The important part when building your own Custom GPT is not getting it perfect first time. Like any system, it improves through use. You test it, notice where it falls short, adjust your instructions, and refine again. This iterative process is where the real value is built. Over time, what starts as a simple Custom GPT can evolve into something far more useful – supporting how you think, how you work, and how you begin to structure tasks so they move forward more easily.
This is where the idea of “cloning yourself” or “building your AI brain” begins – each Custom GPT capturing a specific way you think, so you’re not starting from scratch every time. You’re not actually creating one perfect AI version of you. You’re building a set of specialised versions of how you think. It’s worth noting that one GPT trying to do everything becomes vague, inconsistent and hard to refine, which is why multiple GPTs are better leveraged to give greater clarity, enhanced focus and better predictability
Why Capability Matters More Than Context
A common mistake is creating Custom GPTs tied to specific projects or businesses. This quickly becomes cluttered. A more effective approach is to build around capability. You need support with:
- thinking strategically
- refining messaging
- structuring ideas
- challenging assumptions
Those needs don’t change from project to project. When your GPTs are built around those capabilities, they become reusable. And much more powerful.
Agents and the Wider AI Landscape
Understanding What Matters Without Getting Lost in It
As you explore further, you will come across terms like “agents” and “agentic systems.” You’ll also see a growing number of tools, platforms, and integrations. It can quickly feel like you’re behind. You’re not.
What an Agent Actually Is
At its simplest, an agent is a system setup where AI can work through multiple steps toward a goal, rather than responding once and stopping. Instead of a single interaction – where you ask, it answers, and the process ends – you define a goal, and it works through the steps to move something forward. For example, instead of asking for research, then summarising it, then turning it into something useful, an agent can handle that sequence as a process.
A helpful way to think about this is that an agent behaves like a team member (more than a mode of operation, it’s a particular setup). It’s not an object – but defines a way of working, aka how Ai should operate. For example, you don’t brief someone on a single task and then start again from scratch each time. You give them a direction, a responsibility, and a clear outcome, and they carry that forward – step by step – checking in where needed. Agents work in a similar way. They don’t just respond; they continue. They take a goal and help move it forward across multiple stages.
This is where things begin to shift for many entrepreneurs. For the first time, it becomes possible to extend how work moves forward without immediately needing to hire. You can begin to structure processes, test ideas, and build momentum with a level of support that previously required a team.
But this is also where a new constraint appears.
Because just like with a real team member, the quality of what happens depends on how clearly the work is defined. If the goal is vague, the output will be inconsistent. If the direction is unclear, progress will stall. Agents don’t remove the need for leadership – they make it more visible. This is often the point where many founders realise they don’t just need more execution -they need clearer thinking. And this is where a strong strategic intrapreneurial partner, becomes valuable. Someone who can step in, help define direction, structure the work properly, and ensure that what is being built actually moves the business forward.
In that sense, agents can help carry the work. But it is still clarity – often shaped with the right support – that determines whether that work leads anywhere meaningful.
Agentic systems don’t remove the need for thinking. They increase the need for direction.
What This Looks Like in Practice: Codex
If you’re using ChatGPT, you may notice a “Codex” tab in your menu. This can feel confusing at first, but it’s actually a helpful example of everything we’ve just described.
Codex is a coding-focused agent built into the ChatGPT ecosystem. Instead of simply answering a question and stopping, it is set up to take on a goal- such as improving a piece of code – and work through it step by step. It can read files, make changes, run tests, and suggest improvements, often completing the task before handing it back for review.
What matters here isn’t the coding itself, but the shift it represents. This is the same underlying artificial intelligence you’ve been using, but now configured differently. Rather than helping you think through a task, it is set up to carry the work forward. In other words, it behaves much more like the “team member” we described earlier – not because it is one, but because of how it has been structured.
For most founders, you won’t be using Codex directly in your day-to-day work. But it’s useful to see it as a working example. It shows what becomes possible when AI is set up not just to respond, but to move something forward across multiple steps. Once you see that, the idea of agents stops being abstract and starts becoming something you can apply – whether in coding, content, or building your own systems.
What Matters (And What Doesn’t Yet)
This is where it’s important to stay grounded. Agentic systems are powerful, but they are still developing. There is a lot of excitement – and a fair amount of exaggeration. They don’t remove the need for thinking. They increase the importance of direction. Because now, instead of doing everything yourself, you are deciding:
- what should happen
- in what order
- and to what standard
For most experts building online, this is not the first thing to master. A strong thinking system will take you much further, much faster.
Where Tools Like ChatGPT and Claude Fit
At the centre of all of this are tools like ChatGPT and Claude by Anthropic. These are your core environments for thinking, writing, and planning. Around them are more specialised tools – design, video, automation, integrations. These can be useful, but they are not the foundation. If your core way of working is unclear, more tools will not fix it. If your core is strong, you won’t need many.
A Simple Glossary: Making Sense of the AI Language
One of the reasons AI can feel more complex than it is, is simply because of language. Terms are used loosely, often interchangeably, and sometimes incorrectly. So before going further, it helps to ground a few of the most common ones.
LLMs (Large Language Models)
This is the foundation of everything you’re using. An LLM is the underlying system that has been trained on vast amounts of text and can generate responses, ideas, and structured outputs. Tools like ChatGPT and Claude are powered by LLMs. You don’t interact with the model directly – you interact with a product built on top of it. A simple way to think about it: The LLM is the engine. The tool you use is the interface.
AI Tools (ChatGPT, Claude, etc.)
These are the environments you actually work in. They take the underlying model and wrap it in a usable interface – adding things like chats, projects, files, and sometimes memory. This is where your day-to-day thinking, writing, and planning happens. For most experts, this is the centre of everything.
Prompts
A prompt is simply the instruction or input you give to the AI. Despite how they’re often presented, prompts are not magic formulas. They are just a way of giving direction. Their effectiveness depends far more on your clarity than on your wording.
Context
This is one of the most important concepts, and one of the least understood. Context is everything the AI needs to understand your situation – what you’re working on, what stage you’re at, what matters, and what constraints exist. Without context, you get generic output. With context, you get relevant thinking.
Custom GPTs
Custom GPTs are pre-defined ways of working. They allow you to set how you want the AI to think, respond, and behave before the conversation begins. Instead of repeating instructions each time, you start from a consistent position. They are most useful when you notice patterns in how you work and want to make those patterns repeatable.
Agents (and Agentic Systems)
Agents are systems where AI can take multiple steps toward a goal, rather than responding once and stopping. Instead of answering a single question, an agent can: gather information, process it, take action and move something forward. “Agentic” simply refers to this ability to act in a more independent, multi-step way. That said, these systems still rely heavily on direction. They don’t remove the need for thinking – they increase the importance of it. Think of agentic systems as goal-driven, adaptive behaviour. Instead of following fixed steps, it can decide what to do next, adapt based on what it finds, iterate and refine outputs. It’s not just following steps – it’s figuring out the steps.
AI Workflows
This is where multiple tools or steps are connected together. For example, something might: capture input, process it through an AI, send the output somewhere else. Workflows are essentially a fixed sequence of steps that run the same way each time – useful when things become repetitive. But they only work well when the underlying thinking is already clear.
The Simple Way to See It All Together
If you step back, the landscape becomes much simpler:
- LLMs are the underlying intelligence
- Tools like ChatGPT and Claude are where you work
- Prompts and context shape the quality of thinking
- Custom GPTs define how you work repeatedly
- Agents and workflows extend what can happen over time
One final point, because the language in this space can be surprisingly confusing. You’ll often hear terms like models, tools, and agents used interchangeably, which makes everything feel more complex than it actually is. A simple way to understand it is this: the model is the intelligence – such as GPT or Claude, the tool is where you use it – like ChatGPT or Claude (for Claude the same name is given to both the model and the interface, which can make things feel confusing), and an agent is what happens when that intelligence is given a goal and allowed to carry work forward. In practice, that feels less like asking for a one-off answer, and more like working with something that helps move the task along step by step.
You don’t need to master all of this. But having a clear understanding removes a lot of unnecessary friction. Because instead of feeling like you’re navigating something complex – you start to see the structure underneath it. And once you can see that structure, it becomes much easier to work with.
The Role Shift: From Doing to Directing
At some point, your role begins to change. Not dramatically. But noticeably.
You spend less time trying to do everything yourself, and more time shaping what should happen. You bring the vision, the experience, and the judgement. AI helps you develop, structure, and move that forward. For experts entering the online space, this is where things begin to unlock. And it’s exactly where the right kind of support matters – not to do the work for you, but to help structure it so it can actually move.
Your challenge has never been knowledge. It’s been translating that knowledge into something structured, something scalable, something that doesn’t rely entirely on you holding everything together. This is where AI becomes genuinely valuable. Not because it replaces you – but because it supports how you think.
The shift isn’t from doing less work. It’s from holding less in your head.
Bringing It All Together
How to use ChatGPT for business effectively
When you step back, the system is simpler than it first appears. You have a place where your thinking lives, a way of structuring your work, different modes of thinking depending on what you are doing, and tools that support that process. Most people only ever experience fragments of this. That’s why AI feels inconsistent. The advantage comes when you begin to see it as a whole – not as separate tools, but as a way of working.
What Super-Users Do Differently
Once you understand how to use ChatGPT for business as a system, not a tool, everything begins to compound. By this point, the difference starts to become clearer. It isn’t that some people are better at using AI. It’s that they work with it differently. And over time, a few simple patterns begin to emerge.
They don’t try to manage everything in their head. They create spaces where their thinking can continue, and they return to them consistently. Instead of starting new conversations each time, they build on what already exists, allowing ideas to evolve rather than reset.
They are careful about structure, but not rigid. They keep a small number of meaningful Projects rather than fragmenting their work across dozens of disconnected spaces. They understand that clarity doesn’t come from organising more – it comes from keeping related thinking together.
They don’t rely on AI to “remember” things for them. Instead, they take responsibility for context. They restate what matters when needed, refine direction as things evolve, and treat clarity as something they actively maintain rather than something the tool provides automatically.
Over time, they also become more intentional about how they use conversations. They recognise when they are thinking strategically, when they are executing, and when they are creating. They don’t mix everything together unnecessarily, because they know that different types of thinking require different levels of focus.
And perhaps most importantly, they return. They come back to the same Projects, the same lines of thinking, the same decisions in progress. Because they understand something most people miss. Once you understand how to use ChatGPT for business as a system, not a tool – the advantage isn’t in what AI can do once. It’s in what it helps you build over time.
Final Thought
Most people are still asking how to use AI. But that’s not the real question. The real question is: What happens when your thinking no longer resets every time you sit down to work? Because that’s the shift. And the people who solve that won’t just move faster. They’ll build better. Quietly, consistently, and with far less friction than everyone else trying to keep up.
AI doesn’t give you an advantage because it works faster. It gives you an advantage because your thinking no longer resets.