ChatGPT wrote my personal branding. Eleven months later, here's the diff.
· 3 min readIn July 2025, one month before building this site, I did the classic exercise with ChatGPT: define my personal brand. Full questionnaire — mission, values, passions — then generation of a signature line, an editorial strategy, and a sample article.
Eleven months later, the site exists, and so does the brandbook. I pulled up the original document and ran the diff.
What ChatGPT had produced
A signature line: “Decoding the world around us to better understand, create, and share.”
An editorial strategy in four pillars (AI, methodology, teaching, posture). An article template with the instruction to “always end with an opening — a question, a possible application, or a doubt deliberately left hanging.” And a complete sample article, structured with 🔹 and 👉 emojis, that opened with:
Sharing an idea isn’t just passing it along. It’s making it understandable, actionable, and sometimes memorable.
At the time, I was pleased. It looked like personal branding. That was exactly the problem.
What survived: all of the substance
The questionnaire held up. Eleven months on:
- The mission “Share. Teach. Discover” became the first line of my manifesto: “I design learning experiences and I build them.”
- The three core values it surfaced (knowledge, innovation, competition/risk) became the site’s three activities: instructional design, AI consulting, game studio.
- Being a multipotentialite, which I described back then as something to justify, became the site’s very architecture. I no longer explain it — I display it.
Why did it hold? Because ChatGPT didn’t produce it. The substance came from my answers to the questionnaire. The AI structured; it invented nothing.
What died: all of the style
The signature line was never used. Anywhere. Not one page, not one bio, not one post. It was replaced by “Intent before the tool” — four words found while writing the manifesto, not while brainstorming taglines.
As for tone, the 2025 sample article now violates just about every rule in my Voice & tone section:
| 2025 (ChatGPT) | 2026 (brandbook) |
|---|---|
| “Always end with an opening” | No cliffhangers |
| ”A doubt deliberately left hanging” | Subject + verb + proof |
| Structural emojis 🔹👉✍️ | Zero decorative emoji |
| ”AI as a cognitive partner” | Show the work, skip the story |
The honest read
ChatGPT did the collection work well and the writing work poorly. Which makes sense. The substance came from me: a well-run questionnaire is a free coaching session. The style came from the average of 2025 LinkedIn posts — and writing like the average is the opposite of a brand.
The other lesson: a voice isn’t generated, it’s subtracted. It took me eleven months of publishing to know what I don’t write. No model could have given me that in July 2025, because it didn’t exist yet.
If you do this exercise with an AI: keep the questionnaire answers, throw away the prose.
The 2025 template recommended to “always end with an opening.” This one ends with a period.
Pierre Touzet
Instructional Designer & Digital Innovation
15 years across IT, education and digital innovation. National Program Director, Gradly creator, CNIL-certified DPO.
Learn more →