AI Discovery Workflow: Skincare Discovery Prompting
Training AI to speak with clarity, care, and brand alignment
As AI becomes part of how people search, learn, and make decisions, I wanted to explore what it would take to build a better discovery experience, especially in a space as personal as skincare. This project is about designing structured prompt workflows that guide models like GPT, Claude, Gemini, and Perplexity to deliver consistent, helpful, and trustworthy product responses.
The system I created builds toward smarter discovery by focusing on tone, intent, and structured refinement.
Role & Scope
Prompt engineering, output tagging, multi-model testing, tone shaping, and implementation strategy.
Tools: GPT-4, Claude, Gemini, Perplexity, Notion, Figma
The Workflow
A 7-stage process for testing, evaluating, and refining AI output across models.
Prompt System
I built a modular library of prompts tailored to user intent, skin type, product preference, and tone.
Example prompt: “You are a skincare advisor trained in dermatology. A user asks about recurring breakouts around their jawline. Respond with possible causes, suggest gentle solutions, and only mention non-comedogenic products. Use a calming, affirming tone.”
Model Testing
Each response was scored across five categories:
Accuracy, Clarity, Empathy, Brand Tone, Actionability
The system allowed me to compare behavior across models and adjust prompts accordingly.
Why This Matters
Search is changing. Instead of static filters, we now have conversations with systems. That shift demands clearer tone, higher context, and a more human approach to design. This workflow explores how AI can support relationship-building in futures.
[scroll down to see the full cheat sheet]
