zainAI Playground Workshop
A full-day, hands-on workshop to move my design team from AI-curious to AI-productive. The bet: you don't talk people out of fear, you get their hands dirty, so I built a day of reps, not lectures, where a broken output still counted as a win.
- role
- Organizer + facilitator: designed and ran it end-to-end
- stack
- AI enablement · Workshop facilitation · Team upskilling
- status
- shipped · in use
How I got here
Porter sent me to a company-sponsored two-day AI course. I came back convinced the gap on my design team wasn't knowledge. It was fear. People were AI-curious but not AI-productive: they'd read the threads, admired the demos, and still hadn't logged into a single tool.
I already ran a recurring team ritual I called Design Day: a full day, every couple of months, to step out of delivery and build a shared skill. And I'd been leaning on AI hard in my own planning (fanning the same brief out to several models and curating the results). So I made the first Design Day chapter an all-day, hands-on AI workshop and titled it "AI Playground: Don't Fear the Future, Design It!"
How I thought about it
The core bet: you don't argue someone out of fear, you get their hands dirty. So I designed the day around reps, not slides. Exploration over perfection, a playground, where a broken output was still a win because they'd actually made the thing.
I wanted everyone to leave holding one framing: AI as a superpower, not a replacement, and taste plus domain knowledge (accessibility, empathy, craft) as the thing that still separates good work from generic output.
The trap in a mixed-fluency room is losing the nervous people in the first ten minutes. So I front-loaded plain-language vocabulary (what a model is, what a wrapper is, what machine learning actually means) before any tool. And I made sure each person walked out with a personal artifact: their own face trained into a model, a song about the team, an avatar that talks in their own voice. You remember the magic you felt, not the one you were shown.
I built the deck myself as an interactive web presentation (dark, brand-violet, "made by a designer") with the agenda framed as an "AI Buffet" and flashcards and a quiz woven in to keep it hands-on rather than a lecture.
What I actually did
The day ran the better part of a full day, late morning into the evening, in a single room, as one long guided session.
Foundations. Plain-English AI vocabulary first: machine learning as the process, a model as the output, a wrapper (like ChatGPT) as the interface. Then the map: Hugging Face as "the GitHub for AI models," and Replicate to run them in the cloud so no one's laptop was the bottleneck.
Train your own face: the hands-on spine. Everyone made accounts, generated access tokens, set up billing, and trained a Flux LoRA model on photos of themselves. This was the centerpiece, and the part where the friction was most real.
Prompting. ChatGPT for prompts and negative prompts: how to constrain an image toward what you actually want instead of fighting the defaults.
Make something personal. Suno to generate songs, including a team anthem; and avatar tools (HeyGen, D-ID, Tedra) to animate a still photo and give it your recorded voice.
Put it to work. Alfred for workflow shortcuts; Claude to break research documents into empathy maps and journey maps; and "Gems," a custom GPT I'd built that role-plays a Porter partner and critiques a screen from that user's point of view.
I didn't hide the failures. Payment-gateway and billing errors stalled several LoRA trainings mid-session, and the honest feedback afterward called the day "overwhelming" as often as "well done." Both were true, and I'd rather log both.
Where it landed
The sentiment arc was the whole point. The room started at fear and friction (anxious about just logging into new tools) and ended at "seamless," "super fun," "amazing." Most people named the spark it gave them more than any one tool.
It became the first of what turned into a real program. The workshop introduced avatar-video and AI-voice tools (HeyGen, D-ID, Tedra), which seeded the team's later video and voice work.
The "Gems" partner-feedback GPT, and a workshop idea for an internal copy-checking tool, both carried forward: a couple of months later the team went on to win Porter's first company-wide AI hackathon with a UX-writing Figma plugin, and the one-off Playground eventually professionalized into a structured, recurring AI-upskilling program for the design team.