Most companies measure AI adoption wrong...

What you’ll learn: How to assess your team's AI fluency and go from screenshot to production code.

Amir MXT, founder of Humblytics, has seen it over and over: a company announces they're "going AI-first," rolls out tools, then tracks logins as proof it's working. Six months later, nothing has changed.

His take: the tools matter far less than change management. In this episode, he shares how he assesses a team's real AI fluency and the workflow his team uses to go from screenshot to production in 4 simple steps.

Tools mentioned in this episode

1. Cursor — A code editor with AI built in

2. Figma / Figma Make — Design tool; Make turns images into editable designs

3. MCP connectors — Bridges that let AI tools talk to other apps

4. GitBook — Documentation tool connected to your code

5. Linear — Project management for software teams

6. Firecrawl — Extracts information from websites for AI to use

Episode TL;DR

• Create a shared context file (style guide, docs, ICP) that agents reference on every build. This is how a 2-person team stays consistent

• Measure lead time from request to production, not token usage. Amir's team ships customer requests same-day

• Screenshot → Figma Make → Cursor MCP gets you from image to working React code in under 5 minutes

Tutorial 1: Assess your team's AI fluency

The four levels:

1. Helper: Uses AI for general tasks (summarizing, fixing typos)

2. Automator: Saves prompts and reuses them for consistent results

3. Augmenter: AI woven into daily work, but human reviews everything

4. Agentic: AI takes initiative and executes tasks proactively

How to use this:

• Ask each person: What AI tools do you use? What for? Where does it fail you?

• Track outcomes, not usage. 

  • For engineering: time from request to production. 

  • For support: time from ticket to resolution.

• Establish one rule: "If you ship it, you own it — even if AI wrote it."

• Help each person move up one level with a specific, concrete goal.

Tutorial 2: Screenshot to working code

Note: This is more technical. Best if you have some dev experience or can pair with a developer.

You'll need: Figma account (free), Cursor, Figma MCP connector set up in Cursor

The workflow:

1. Screenshot any UI you want to recreate

2. Upload to Figma Make with prompt: "Recreate this UI"

3. In Cursor, connect to your Figma file and ask it to implement the design

4. Refine with natural language: "Make the button bigger," "Add more whitespace"

Pro tip: Add a style guide file to your project so the AI stays consistent with your brand.

When building gets cheap and fast, the advantage goes to teams who pick the right work and take responsibility for what they ship.

👉 Subscribe to This New Way: YouTube Channel

📩 Was this email forwarded to you? Subscribe here

📱Find us on Instagram and TikTok: @ This New Way

Until next time,

Aydin Mirzaee, CEO at Fellow.ai and host of This New Way podcast