Pro Website Creators AI Discussion – Week 11: Training AI to Sound Like You and Automate Your Workflow
This week focused on something practical: teaching AI to write in your voice and building smarter prompts that save time. Dave walked the team through lessons from Dan Martell’s “CEO Four-Day Challenge” and showed how tone documents, system prompts, and custom GPTs can speed up real client work.
What We Tried This Week
Building Your Personal Writing Style Guide
Dave shared how he gathered 35 samples of his writing and fed them into ChatGPT to generate a tone document. The model distilled his style into a warm, conversational, encouraging voice with clear structure and practical phrasing.
When he tested it on a blog post, the difference was obvious — the generic version sounded robotic, while the tone-trained version felt human and closer to Dave’s natural voice. Even an AI detector showed a major drop from 97 percent AI to 66 percent AI after applying his tone.
Turning Prompts Into Systems
Next, Dave demonstrated how he used that tone guide to build a system prompt for PWC’s client survey and review emails. The prompt asks three quick setup questions — client name, company name, and which template to use — and then generates a complete message in Dave’s voice.
He tested it live with Kenneth’s WebDev Studio responses, and the custom GPT pulled the correct template, filled everything in, and wrote a follow-up based on the client’s feedback. This workflow now removes repetition and produces consistent, personalized communication.
Projects vs. Custom GPTs: When to Use Each
The team discussed when to store work inside a Project versus when to build a standalone custom GPT. Projects are ideal for long-term memory — like blog creation, SEO reports, or newsletter summaries. Custom GPTs work best for quick automations where history doesn’t matter, such as generating survey emails or structured templates.
Using AI as a Personal Tutor
Kenneth showed how ChatGPT’s new “Study and Learn” feature is helping him relearn Portuguese. It built a simple lesson plan, asked follow-up questions, and adapted based on his comfort level.
It was a reminder that AI can coach you in almost anything if you give it context and a clear goal.
News Highlights
ChatGPT Group Chats
OpenAI rolled out pilot multi-person chats. It’s limited for now but shows how collaborative AI will soon feel more like a shared workspace.
NotebookLM Deep Research Tools
Google upgraded NotebookLM with file-based deep research, similar to ChatGPT Projects. It’s becoming a strong companion for document-heavy work.
AI-Assisted Cyberattacks
A report surfaced showing Chinese hackers using Claude to automate nearly 90 percent of a cyberattack. It’s a sign that cybersecurity defenses must adapt quickly.
Zapier Multi-Agent Workflow
A data manager trained separate agents on multiple research reports, then built a classifier that routed questions to the correct document. Marketing and leadership now get instant answers with citations.
Google’s New AI Shopping Tools
Google AI Mode can now track prices, compare retailers, surface reviews, and even automate purchases when items drop below your budget.
Key Takeaways
AI becomes far more useful when you train it with your own voice and repeatable systems. A tone guide plus a well-built system prompt can automate hours of communication. Tools like Projects, NotebookLM, and Zapier workflows are opening doors to smarter research, cleaner SOPs, and more personalized learning.
