Read time: under 8 minutes
Hey there, Product Creator. 👋 I was on vacation in Sardinia, a little Italian island, these last few weeks. My favorite place I visited was called Capo Testa, a peninsula off the most northern part of the island filled with moon-shaped granite rocks.
Seeing everything that's been launched in the world of AI gives me a bit of FOMO. This feeling is normal, even when you're not disconnected and on vacation. There are so many innovations emerging, it would be impossible to be an early adopter of each and every one. Let’s dig into the highlights in this edition of the Newsletter.
In this edition:
Creator Spotlight: An analyst who built a daily Fibonacci-inspired puzzle game
Tech Tools: ChatGPT’s morning briefings, Meta’s AI video feed, and AirPods with live translation
Tutorial: Tokenization, explained with LEGO bricks
CREATOR SPOTLIGHT 🎤
Meet Michael Duarte, a Research Analyst at USCIS who built Knotilus, a daily puzzle game that hides Fibonacci-like sequences in a field of numbers. Michael isn’t a PM or engineer by trade—this was a spare-time project powered by curiosity and help from AI. By framing math puzzles as a game, he made them approachable and fun. Think Sudoku with a mathematical twist, delivered one bite-sized challenge per day.
Michael leaned on GPT-5 as a collaborator. It helped him debug endpoints and refine schemas, but also polish copy and test messaging. He put in ~80–100 hours over evenings and weekends—the first couple of weeks on generating unique puzzles, and the remaining month on UI and storage. His takeaway: AI won’t build the product for you, but it can accelerate both your systems and your storytelling. Check it out.
Tools used: Vue, Flask, IndexedDB/Dexie.js, PostgreSQL, GPT-5
Time to MVP: ~80–100 hours over 2 months of nights and weekends
Top tip: Treat AI like a collaborator. It won’t build the product for you, but it will accelerate your thinking and help you see paths you may have missed.
💡 Building something yourself? Share here to be featured in an upcoming newsletter!
TRENDING TECH TOOLS 🔨
Click here to watch OpenAI’s ChatGPT Plus announcement. Source: OpenAI
1. ChatGPT’s new morning briefing (Pulse): OpenAI introduced Pulse, a feature that generates overnight “briefing cards” with your calendar, inbox, and curated news. Imagine waking up to a personalized dashboard covering meetings, emails, and world events—all written in natural language. For now, it’s rolling out to Pro users, but Plus and free users are expected to get access later. If you’re someone who spends the first hour of your day piecing together what’s happening, this could be the first real replacement for the morning scroll.
2. Meta’s AI video feed (Vibes): Meta launched Vibes, a short-form AI-generated video feed powered by Midjourney and Black Forest Labs. You can browse, remix, and share clips directly into Instagram or Facebook. After witnessing amateurs throwing together obviously AI-generated videos on Instagram, I feel like this was inevitable. But it also raises new ethical questions: how do we separate authentic creativity from algorithmically generated slop, and do we even care?
3. Apple’s AirPods Pro 3 with live translation: Apple’s upcoming AirPods upgrade goes beyond audio. At $249, these earbuds will translate conversations live, track your heart rate, and cancel noise better than before. For PMs and travelers, this is more than a gimmick. Imagine testing a product with users in another language and having real-time translation whispered in your ear. Apple is positioning AirPods as an AI-first wearable—not just a pair of headphones. Check them out here.
LEARN A NEW TERM 🧠
Image generated with ChatGPT.
Tokenization → the building blocks of language models. Large Language Models (LLMs) don’t read words the way we do. They break everything down into tokens—chunks of text that can be as small as one character or as long as one word.
Think of tokens like LEGO bricks. Each brick on its own is simple, but when connected in sequence, they can build something complex—like sentences, code, or stories. “Banana” might be one brick, but “Bananasplit” might be two.
Why it matters for PMs: Tokenization affects performance and cost. An API call priced “per 1,000 tokens” isn’t abstract—it’s literally how much text the model can process. If your product is generating long prompts or storing chat histories, understanding tokens helps you estimate costs and optimize inputs.
FROM THE ARCHIVES 📁
Missed this carousel? I kicked off my series “APIs for Product Managers” with a Friends-themed carousel.
Ordering a latte at Central Perk is the perfect analogy: you make a request (“one latte, please”), the barista checks authentication (your VIP card), the request is routed to the right endpoint (espresso machine vs blender), and then makes its way back to the client (you, the customer). Check out the full carousel here.
PROMPT PLAYGROUND 🤖
This edition’s featured prompt: Mini Puzzle Builder.
Inspired by Knotilus? Here’s a copy-&-paste prompt designed to help you get a functional micro-puzzle app up in around 10 hours:
You are my technical guide. Help me build a micro-puzzle app (web or mobile) in ~10 hours.
It should deliver one small puzzle per day.
Step 1: suggest 2–3 puzzle types (e.g., mini sequence, pattern match, small grid) that are simple but engaging.
Step 2: pick one type and define the game loop: how users start, how they solve, how they submit.
Step 3: generate pseudocode that can produce a new puzzle daily (with uniqueness, minimal randomness).
Step 4: recommend a minimal tech stack (frontend + backend + database) that’s free/cheap to host.
Step 5: propose a simple data model for puzzles, user progress, and scores.
Step 6: write starter code snippet (in JS or Python) for the puzzle generation and API endpoint.
Step 7: draft user onboarding copy (2–3 sentences) that explains the puzzle in friendly terms.
Constraints:
Must be buildable in ~10 hours by one person
No need for mobile app stores (web or PWA is fine)
Use serverless or low-ops hosting (e.g., Vercel, Netlify, Firebase)
Keep UI minimal (inputs, grid, submit button)
Output: A step-by-step project plan with sample code, a database schema, a recommended stack, and user-facing copy that you can immediately start building from.JACKIE’S WEEKLY PICKS 💖
✅ Something to read: 10 Product management trends 2025 — Airtable’s take on how PMs will evolve in the AI era, from embracing agentic workflows to rethinking collaboration, tooling, and measurement.
✅ Something to watch: The Deepest Breath — my all-time favorite freediving documentary, remembered while I was in Sardinia practicing my “two-feet-underwater” dives.
✅ Something to listen to: Hamel Husain & Shreya Shankar on Lenny’s podcast — why AI evals are the hottest new PM skill, explained in plain language for anyone who’s ever wondered “what the heck are evals?”
NEW PM ROLES 💼
Google opened applications for the Associate Product Manager Internship for summer 2026, the most competitive PM internship in the industry.
Baylor Scott & White is hiring a Product Manager II to help improve scheduling through AI agents in its hospital system.
Siena AI is hiring an Associate Product Manager to help build AI agents for businesses at an early-stage startup.
SHARE THE LOVE 💌
Like what you read in this week’s edition?
Copy & paste to refer a friend!
https://jackie-the-product-girl.beehiiv.com/subscribe