Gamma’s 3-Step AI Workflow for Global Feedback, Art Direction, and Hiring
By: Zach Leach Host: Claire Vo Source: How I AI (ChatPRD) Type: podcast
Summary
How I AI episode with Zach Leach, Head of Design at Gamma (AI-native presentation tool, ~30 person team, 60% international users). Three workflows showing how a small team uses AI to operate at global scale: (1) Multilingual feedback analysis — exported 550 user responses about AI image editor feature, uploaded to ChatGPT Deep Research with product-focused prompt, got comprehensive translated/categorized analysis in 19 minutes (vs. cherry-picking 20 English responses before), pasted output directly into Gamma for instant shareable presentation with charts that drove real roadmap decisions; (2) Scalable art direction with Midjourney — created brand “kit” using Style References (–sref) so anyone can generate on-brand images, rapid creative iteration following intuition through happy accidents (random bird → final concept), final polish with Replicate for high-quality background removal → Figma; (3) Standardized job descriptions with Claude Projects — uploaded best existing JDs as few-shot examples, any hiring manager generates on-brand postings with simple prompts, paste into Gamma template for instant careers page. Core insight: AI as personal data researcher, art department, and hiring coordinator frees the design team to focus on craft, detail, and delight.
Key Ideas Extracted
- ChatGPT Deep Research for feedback at scale: Upload hundreds of multilingual responses, get a product-focused analysis that translates, categorizes, and surfaces themes — 19 minutes vs. the old approach of cherry-picking a few English responses
- Prompt specificity matters: Basic ChatGPT prompt just wrote a Python keyword-matching script; specifying Deep Research and giving a detailed product-focused prompt got genuinely insightful nuanced analysis
- Midjourney Style References as brand kit: Use –sref to create reusable brand styles so anyone at the company can generate on-brand images without design expertise
- Follow creative intuition with zero friction: Rapid iteration through hundreds of variations in Midjourney — happy accidents (random bird in an apple image) spark new creative directions that would take weeks with traditional agencies
- Replicate for final polish: ML models for specific tasks like high-quality background removal — seconds vs. manual pen tool tracing, adds that extra layer of craft
- Claude Projects for organizational consistency: Few-shot prompting with best existing examples lets any hiring manager generate on-brand job descriptions — saves time, reinforces quality, and is reusable team-wide
- Feedback → presentation pipeline: ChatGPT analysis → Gamma auto-generates presentation with charts → immediate basis for product/engineering roadmap conversations
- AI enables lean teams at global scale: 30-person team serving international users without sacrificing quality on research, brand, or hiring
Notes
- Published Jun 9, 2025 on How I AI (ChatPRD). ~9 min read. (Note: filename says 2025-08-18 but actual publication date is Jun 9, 2025)
- Sponsors: WorkOS, Retool
- Zach Leach background: Head of Design at Gamma
- Tools: ChatGPT (Deep Research), Midjourney (–sref Style References), Replicate (background removal), Claude Projects (few-shot JDs), Gamma (presentations), Figma
- Gamma stats: ~30 person team, 60% international user base
- Feature context: AI image editing feature feedback analysis
- Three companion workflow guides published Jan 8, 2026
- Cross-references: Feedback synthesis, lean team scaling, AI art direction
Raw Content
Re-scraped from ChatPRD 2026-02-16. Full article content captured in Summary and Key Ideas above.