The AI Tool Stack I Use to Train 3,000 Professionals a Year
People ask me which AI tool is "the best." After training over 3,000 professionals as an AI trainer in Hong Kong, my answer is always the same: it depends on the room.
The tool stack I use changes for every workshop. A bank gets a different toolkit than a jewelry company. An engineering firm gets different tools than a university. But after two years of refining this across industries, I've settled on a core stack with modular additions.
The Core Stack (Every Workshop)
Microsoft Copilot -- the default starting point for any enterprise engagement. It's inside the security boundary, requires no new accounts, and integrates with the tools people already use daily. I use this for: email drafting, meeting summaries, document generation, basic research.
ChatGPT / Gemini -- for tasks that need more horsepower than Copilot provides. Complex analysis, creative brainstorming, multi-step reasoning. I teach people to use these as the "thinking partner" tool when Copilot's built-in limitations surface.
NotebookLM -- Google's underrated gem. I use this for information consolidation and research synthesis. Upload 50 pages of meeting notes, and NotebookLM creates a structured summary with citations. At CTF's design workshop, teams used it to go from scattered market research to a coherent design brief in under 20 minutes.
Industry-Specific Additions
For creative teams (design, marketing):
- Midjourney -- mood boards, style exploration, concept visualization
- Nano Banana Pro (Google AI Studio) -- precision work, especially for product design. This produces the best results for jewelry, fashion, and product visualization
- Lovable -- rapid landing page prototyping. At CTF, teams went from concept to functional prototype in 30 minutes
For engineering and technical teams:
- Copilot in VS Code -- for teams with any coding component
- Claude -- for technical writing, documentation, and complex reasoning
- Perplexity -- for technical research with citations (engineers demand sources)
For executives and management:
- ChatGPT Advanced Data Analysis -- for spreadsheet work and data visualization
- Gamma or Beautiful.ai -- for presentation generation
- Calendar/email AI integrations -- for workflow automation
What I Don't Teach
I deliberately exclude tools that are:
1. Too volatile. If a tool changes its pricing or features monthly, I don't build a workshop around it. My toolkit includes tools that have been stable for at least 6 months.
2. Not enterprise-safe. Any tool that doesn't have clear data handling policies gets excluded. When I'm working with banking professionals at scale, I can't afford ambiguity about where data goes.
3. Redundant. I'd rather teach 3 tools deeply than 10 tools superficially. Most workflows can be covered with Copilot + one general-purpose LLM + one specialized tool.
The Meta-Lesson
The most important thing I teach isn't any specific tool. It's the framework for evaluating any new AI tool that appears next month:
- What does it do that my current tools don't?
- Is it enterprise-safe for my organization's data?
- Will I use it daily, or is it a novelty?
If a tool doesn't clear all three, it's a distraction. The AI landscape changes weekly. The skill of evaluating tools is more durable than knowledge of any specific tool.
As an AI trainer in Hong Kong, my job is to send people back to their desks with tools they'll actually use tomorrow -- not a list of 50 apps they'll forget by Friday.
