Geargina Tan is a Singapore-based AI workshop facilitator who gets business teams using Claude and ChatGPT for real work. She is co-founder and COO of WTFox.ai and writes as Coach Gina. Two ways in. The Claude Masterclass 101 as an open cohort, or a private Claude session built on your team's own documents.
Last updated 9 September 2026
I run WTFox.ai on Claude. Not as a demo, as the thing the company actually operates on, which is the only reason I've got an opinion worth listening to here.
Four pieces do the work, and they stack in an order a beginner can follow. Projects hold your context, so a chat starts warm instead of cold every morning. Connectors reach into Gmail and Drive with deliberately narrow access, Drive read-only and Gmail read and draft. Skills make one task repeatable, written in plain instructions with no code anywhere in the file. And Cowork takes a job end to end while you go and do something else.
Claude or ChatGPT? I'm not going to hand you a benchmark table, and anyone who does is measuring something your team will never notice. Both are good. Claude is where I start a non-technical room, because those four pieces have an order to them and people can see themselves climbing it. ChatGPT is broader and more of your team have already tried it. That's a teaching preference, not a verdict, and I teach both.
Pick the model, set the effort, and brief it like a new hire on their first day rather than a search box. We run the same question twice in front of the room, cold then briefed, and the difference does the arguing for me.
Where most people stopA Project holds your instructions, your documents and your tools in one place, so it answers warm. Connectors give it hands, with narrow access and your login. This is the step where people stop starting from zero every morning.
Where it starts payingA skill teaches it once so the task runs the same way every time, and Cowork finishes a job end to end. The habit worth installing is files plus outcome, not steps, then reviewing what comes back the way you'd review a new hire's first draft.
Where the hours come back3 hrs, hands-on the whole way, capped at 30 people and built for a room that is new to AI. It runs with Cogentic AI Partners, who own the marketing and registration. Bring a laptop, a Claude account logged in before you arrive, and one real piece of your own work.
2 evening cohorts ran on 20 and 21 Aug 2026 in Singapore with 21 paying attendees. The room worked through 4 builds. A Claude Project around their own work, their tools connected, a first custom skill written, and one real task handed over to Cowork before the night ended.
The numbers, with the caveat attached. 91% of survey respondents rated the pacing just right, which was 10 of 11. Self-rated confidence in setting up a Claude Project rose 48%. That came from an anonymous survey before and after, 19 responses before and 11 after, and they are not the same 11 people, so treat it as post-session feedback rather than a study.
Registration is with Cogentic AI Partners. The 68-second trailer walks through the night if you want to see the shape of it first. Level 2, on Claude Code and building a second brain, is planned for Oct 2026 as well, with the date still open.
Same ladder, your material. 2 hrs to a full day, in person in Singapore or online, working on the documents and the tasks your team already has open. A 2 hr session gets a room to chat and context. 3 hrs is the format that reaches a skill and a delegated task. A full day gives the build blocks room to breathe.
The run sheet, the three formats side by side, the survey and the honest bit about what went wrong all sit on the workshops hub. Pricing is scoped per team.
Everything taught in the room is something I run myself. The dashboard below is my own, demoed live in sessions rather than screenshotted for a slide.
The full write-up of the August cohorts is at the Claude Masterclass 101 case study. The concept underneath connectors gets its own explainer on the blog: API vs MCP, explained simply. And if you're still deciding between tools, why "just use ChatGPT" is bad advice.
Whichever one you already pay for. If it's an open choice, I start non-technical rooms on Claude, because Projects, connectors, skills and agentic work stack in an order people can follow without touching code. ChatGPT has the broader ecosystem and more of your team will have tried it already. That's a preference formed from teaching, not a benchmark, and I teach both.
For the Masterclass, yes. The pre-work note asks everyone to arrive with a paid Claude plan, confirmed in the pre-work note, and already logged in, because the agentic block needs one. For a private team session it depends how far up the ladder you want to go, so we confirm plan requirements on the scoping call before anyone's asked to pay for anything.
4 things, in the 3 hr Masterclass format. A Claude Project holding your own documents and instructions. Your tools connected. One custom skill written, and one real task handed over to Cowork end to end. That's what the Aug 2026 cohorts built. Shorter formats stop earlier on the ladder, and the scoping call decides how far we go.