Last week in a Hangzhou café, I sent the same proposal to three clients

Xiaolin is a freelance brand consultant. On Monday morning, she met me at a café by West Lake in Hangzhou and said a client had read her AI-written service introduction and replied: “This looks like you wrote it yourself, right?”Nothing was wrong. It was just too template-like. She was frustrated too: “I got stuck here too. I spent three hours changing the wording, but all I did was swap a few synonyms.” When a client can tell at a glance that the content skipped real thought, I’d resist arguing with the AI and start by locating the actual problem.

AI isn’t ghostwriting here; it’s helping organize the material

The core tool is actually ordinary: any AI assistant that can handle long text. It cannot decide what a client truly cares about, but it is well suited to turning interview notes, project context, and past conversations into a clear structure.

I know a designer, Chen Mo, who reviews projects every Friday at his studio in Suzhou. He puts client questionnaires, event notes, and his own thoughts into one document, then asks AI to list the needs people keep mentioning. After that, he checks each item himself: which requirements the client stated explicitly, and which are only his assumptions. He uses that list to draft a reply instead of asking AI to write a polished letter from scratch. Chen Mo says, “Now I use it to save organizing time, not judgment.” That’s the role that has worked for him.

What it costs to replicate this today

For money, start with a monthly budget of RMB 0—100: most entry-level chat plans should be enough, and I’d only pay more if usage requires it rather than buying a premium plan immediately. For time, expect about 30—60 minutes for the first setup and roughly 15—30 minutes per use afterward. The technical barrier is low: being able to copy and paste text and check whether AI has invented anything is enough. I don’t write code, and I don’t need to understand complicated interface settings.

The first step is simple: open the AI chat page I already use, start a new conversation, paste in a client requirement or my project notes, and enter: “Please list the facts in the source text, my assumptions, and the information still missing. Don’t write the proposal for me.” I’d check whether it separates facts from speculation before deciding whether to continue. Not everyone needs this tool; if I can already organize material quickly, there’s no urgency to try it.

Here’s how I’d use it at each stage

If I’m just starting out: I’d practice with my own introduction page first. I’d have AI list the five questions clients are most likely to ask, then add real experiences and case studies myself. I wouldn’t chase polished writing yet.

If I already have 1—2 clients: I’d spend the first ten minutes before each conversation organizing the requirements. I’d ask AI only three questions: What does the client really want? Which details contradict each other? What evidence am I still missing? I’d choose the approach; AI would only help organize the thinking.

If I’m scaling up: I’d have the team keep one standard client-information template, while requiring everyone to label the source of each detail. AI could organize the material, but it couldn’t skip that step and generate commitments directly. Once the workflow is running smoothly, I’d consider more automation; there’s no need to build a complex system from day one.

If I wanted, I could try this tonight with a single real client message. The goal isn’t prettier writing. It’s letting the other person see that I genuinely understood their situation.