W02 · Scope, Systems & Sanity

The working relationship with AI is part of what we cover in Workshop 2 — how to protect your time, your voice, and your judgment when working with tools that move fast.

See the workshops →
PROCESSAI COLLABORATIONPROFESSIONAL PRACTICE

How to work with AI without losing your mind (or your voice).

The brief, the mistakes worth making, and why every new project is a new relationship.

This one is harder to illustrate with a before-and-after image. It is about posture — the way you show up to a working relationship — and how that posture determines what you get out of it.

I came to AI tools with a background in brand identity, illustration, and design direction — work that has always been about knowing what you want and communicating it precisely enough that someone else can help you get there. That skill transferred directly. What I didn't have was patience for the learning curve, and what I underestimated was how much the quality of my output depended on the quality of my input — not just the prompt, but the state I was in when I wrote it.

I started using ChatGPT in mid-2025 for research, drafting, and problem-solving. I began working with Manus — a more autonomous AI that can build, iterate, and deploy — on March 29, 2026. In that time I have made most of the mistakes that are worth making.

The brief is not just for the AI — it's for you

I have given vague prompts and been frustrated by vague results. I have accepted output I didn't like because I didn't know how to ask for something different. I have had the AI change things I didn't ask it to change — and had to spend time undoing what it did. I have also had moments where the collaboration produced something I couldn't have produced alone, faster than I expected, in a direction I hadn't anticipated.

The difference between those two experiences is not the AI. It is the brief.

"A brief does not make the AI smarter. It makes you clearer."

When you sit down to write a brief — even a short one — you are forced to articulate what you want, why it matters, and what constraints apply. That articulation is the work. The AI is just the collaborator who receives it.

The AI Brief Builder in Workshop 1 is structured around five questions:

01

Who are you, and how do you work?

Your background, your working style, your voice. What the AI needs to know about you before it can help you.

02

What skills are you bringing to the project, and where do you need help from AI?

Be specific about what you own and what you're delegating. The AI works best when it knows its role.

03

What are you trying to accomplish in this session?

Specific, bounded, achievable in one sitting.

04

What do you already have, and what are you starting from?

Context the AI cannot guess.

05

What does a good result look like — and what would make it wrong?

The clearest question most people skip.

These are not AI questions. They are the questions you would ask before any creative collaboration. The AI just makes the cost of skipping them visible immediately.

The state you're in matters

When I started building the workshop series, I was already spent. I needed rest. I needed to step back and write my brief properly — but I didn't. I pushed through, and my prompting suffered for it. Lack of sleep, lack of nourishment, lack of fresh air — the relationship started on the wrong foot, and the output reflected that.

This connects directly to what Workshop 2 — Scope, Systems & Sanity — is about. Running a freelance practice without burning out is not just about managing your workload. It is about recognising when you are not in the right state to do your best work, and building the systems that protect you from yourself when you're not. The same principle applies to working with AI. You cannot brief clearly when you are depleted. The tool will not save you from that.

Every new project is a new relationship

Here is something that surprised me, and that I now consider essential knowledge for anyone working with AI: every time you start a new project, you are starting a new relationship. The AI does not remember you. It does not know your voice, your preferences, your history, or your standards.

This was genuinely frustrating at first. I had built a working relationship over weeks, and then started a new task and found myself back at the beginning. The tone was more formal. The suggestions were more generic. The responses didn't feel like they knew me.

It is not unlike when a new account manager takes over your file at an insurance brokerage, or a client success contact changes mid-project. The new person is not inept. The relationship is not broken. But the context is gone — and rebuilding it takes time and deliberate effort. The brief is how you give that context back, immediately, without having to rebuild it from scratch in conversation.

A good brief is not just a task description; it is a context document. It tells the AI who you are, how you work, what you care about, and what you've already decided. The more of that you give at the start, the faster the working relationship calibrates.

By the end of building this site, something had shifted. The responses were more direct. The tone was closer to mine. Not because the AI had changed — but because it knew more about how I work, what I care about, and when to push back. That is not a feature. That is what happens when you invest in the brief.

The mistakes worth learning from

Accepting the first output.

The first output is a starting point, not a result. The AI is making its best guess based on what you gave it. If you accept the first output, you are accepting its guess rather than your intention. The illustration case study is the clearest example: seven versions, each one closer to what I actually wanted, none of which would have existed if I had stopped at version one.

Not saving as you go.

The AI has no memory. If you don't save what it produced, it is gone. This applies to images, text, code, anything. Screenshot, copy, download — before you ask for the next version.

Changing things that weren't broken.

This one goes both ways. I have asked the AI to improve something and had it change things I didn't ask it to touch. The brief is the protection against this: the more specific you are about what you want changed and what you want left alone, the less likely you are to lose something that was working.

Treating it like a search engine.

The AI works best when you treat it like a skilled contractor who needs a proper brief — not a search engine you type fragments into. A contractor who receives a clear brief can push back, ask clarifying questions, and bring their own judgment to the work. A search engine just returns results.

What this means for your practice

The working relationship with AI is a skill. It is learnable, and it improves with practice. But the practice is not about learning prompts — it is about learning to articulate what you want clearly enough that someone (or something) else can help you get there. That is a communication skill. It is the same skill that makes you better at briefing clients, managing collaborators, and giving feedback.

This is what the workshop series is built on. Not AI as a shortcut, but AI as a mirror — a tool that makes your thinking visible and your intentions testable.

Scope, Systems & Sanity

Protect your time. Keep your voice.

This case study is part of what we cover in Workshop 2 — how to set up a working relationship with AI that protects your judgment, your time, and your professional voice.

← Back to the series
Case Study 02All case studiesCase Study 04