Ten years. Six companies. One newsletter that actually teaches you something. Peter Yang shipped features used by millions, then built a media platform read by 148,000 people who want to know how he did it.
Before "raid" was a verb people used casually in gaming communities, Peter Yang noticed something. Streamers on Twitch were doing this weird thing at the end of broadcasts - manually pasting links in chat to redirect their audience to someone else's stream. It was messy, organic, and completely ad hoc. Yang saw a product in the behavior, built the feature, and watched it grow 4x in a single year to become the number-one creator-to-creator discovery channel on the platform. That is how Peter Yang works: he watches what people actually do, then builds the thing that makes it possible to do it better.
Now he runs Behind the Craft, a weekly newsletter and podcast at the intersection of AI, product management, and creator strategy. With 148,000+ subscribers - and 210,000+ followers across his Substack profile - Yang has built one of the most genuinely practical media properties in the tech learning space. Not the flashiest. Not the most viral. The one people forward to their PM colleagues because it answers questions they actually have.
His current day job is Principal Product Lead at Roblox, where he builds creator tools and analytics dashboards that have already doubled the share of creators who use data to drive their work. Yang is the rare PM who thinks like a creator and creates like a PM - and 148,000 people read his newsletter each week partly because that combination is rarer than it should be.
Default states, edge cases, and good copy - these details are what separates a great product from slop. You have to give a damn about the tiniest details to ship something you can be proud of.
- Peter YangThe path here was not a straight line. Yang spent roughly three years trying to break into product management, a period he describes as having nearly made him give up entirely. He came from Applied Mathematics-Economics at Brown, went through investment banking and consulting, found himself in product marketing, and slowly ground his way in. That experience is not incidental to who he is as an educator. When he writes about PM careers, he writes like someone who remembers what it felt like to be outside the room.
He eventually landed at Facebook where he grew Facebook Live from zero to millions of viewers. Then Twitter. Then Twitch, where the Raids story happened. Then Credit Karma, where he led a credit cards vertical serving 30 million members and ran PM training for over 100 employees internally. Then Reddit, where he launched Reddit Talk - a live audio product that scaled to millions of users before being discontinued when it no longer fit the platform's strategy. He tells that story too, without euphemism.
There is a thread through all of it: Yang is drawn to social and creator-facing products. Live video. Live audio. Raids. Creator analytics. He is fascinated by the mechanics of how people build audiences and connect with each other online - and has spent a career making those mechanics work better for the people who rely on them.
Weekly AI tutorials and interviews with the people actually building the future. Practical by design. Hype by none.
The story of Twitch Raids is a small masterclass in how good product managers think. Yang wasn't handed a feature brief. He noticed an organic behavior - streamers typing channel names into chat to direct their audiences - and recognized the shape of a product underneath it. The question wasn't "should we build this?" The question was "why hasn't this been built yet?"
He built Raids, and within a year it had grown 4x to become the top creator-to-creator discovery channel on Twitch. That feature now lives in the DNA of how live streaming works. It normalized the idea that an ending stream could be a beginning for another creator - a generosity mechanic baked directly into the product.
Years later, that same instinct drives his newsletter. He watches what builders and creators actually struggle with - AI tool overload, PM career anxiety, the pressure to go viral vs. build trust - and converts those observed struggles into actionable tutorials. The medium changed. The approach didn't.
The biggest problem with the creator economy in my opinion is the extreme power law - the top 1% of creators make 99% of the income.
- Peter YangYang's career didn't launch - it was earned. He came out of Brown's Applied Math-Economics program, detoured through investment banking and consulting, and spent three years trying to get into product management before it finally clicked. That struggle made him a better teacher, and it made the book he would eventually write more honest than most.
Each stop in his career maps to a specific chapter of the creator and social product story. Facebook Live during the live video boom. Twitch Raids during the rise of streaming culture. Reddit Talk during the podcast-goes-live era. Roblox creator tools as user-generated content platforms become the dominant entertainment medium for younger generations. Yang has been inside every major moment in social and creator product over the past decade.
Yang started his newsletter while working in fintech, writing about his experiences building creator products. It was originally called "Creator Economy by Peter Yang" - a fairly accurate if uninspired title. In 2024 it became Behind the Craft, and the mission sharpened: extremely practical AI tutorials and interviews for busy people.
That word "practical" does a lot of work. There is no shortage of AI newsletters listing tools or speculating about the future. Yang's newsletter compares ChatGPT, Claude, and Gemini head-to-head, ranks 24 AI tools from essential to forgettable, and gives you step-by-step workflows for using Cursor, Replit, and Claude for actual product work. The tone is tutorial-first, theory-second. If it can't be acted on this week, it usually isn't in the newsletter.
The podcast extends this with 103+ interviews with people actually building the AI era: executives from Anthropic, Google, Figma, Cursor, Replit, Perplexity. Not pundits - practitioners. Yang has interviewed Figma CEO Dylan Field and dozens of AI leaders who rarely do press. The caliber of guests reflects the trust he's built by not wasting his audience's time.
Paid subscribers ($20/month or $150/year) get 27+ AI prompt templates, access to a 700-member Slack community, his Technical Foundations for PMs course (valued at $899), and tool credits. The community is active in a way that most newsletter communities aren't, because Yang spends roughly half his time engaging directly with it.
Weekly AI tutorials and product leadership content. Practical, tactical, ranked #25 in Substack Technology paid tier globally.
103+ episodes with AI and product leaders from Anthropic, Google, Figma, Cursor, Replit, Perplexity. Published every Sunday.
8-day bootcamp. Covers AI workflows, rapid prototyping with Claude, v0, and Cursor. 4.6/5 rating (92 reviews). $899.
2-day cohort for 1-5 year PMs. The only 5-star PM course on Maven in this category. $599.
Published 2020. Amazon bestseller. "The book I wish I had when I got started." Covers PM principles and breaking into PM.
Yang's newsletter started as a creator economy publication. Today it functions primarily as an AI education platform. The transition wasn't a rebrand - it was a reckoning with where the action actually is.
When ChatGPT launched, Yang called it the most impressive piece of consumer technology he'd ever encountered. He began integrating AI tools into his own PM workflows at Roblox, documenting what worked, what was overrated, and what was genuinely transformative. His readers noticed. The content they engaged with most wasn't creator monetization strategy - it was "which AI tool should I actually use and why."
So he leaned in. The newsletter became a testing lab: ChatGPT vs. Claude vs. Gemini, head-to-head and use-case specific. 24 AI tools ranked from essential to forgettable. Tutorials on using Cursor and Replit to prototype products rapidly. Interviews with founders at every major AI company. And throughout all of it, the same underlying question that has always driven his work: what do practitioners actually need to know?
He now says he uses Claude more than ChatGPT for most writing tasks - a specific, quotable preference that signals the kind of concrete recommendations his audience has come to expect. Not "it depends." Which one, and when.
I now use Claude more than ChatGPT (even with GPT-4o) because it's superior at writing.
- Peter YangNobody cares about your FAANG pedigree or AI product certificate. Hire high agency people who have built great side projects or demonstrated proof of work.
Less than 10% of PMs actually dogfood their product on a weekly basis. Use your product like a first-time user and write a friction log of how annoying the experience is.
The risk of viral tweets like this is that it attracts an audience that follows you for the laughs instead of for your expertise.
A PM's job is to empower their team to build products that solve customer problems and make business impact.
Default states, edge cases, and good copy - these details are what separates a great product from slop. You have to give a damn about the tiniest details.
The biggest problem with the creator economy is the extreme power law - the top 1% of creators make 99% of the income.
A few things set Yang apart from the broader PM-turned-creator crowd, and they're not hard to identify if you read him long enough.
He's anti-credentialist in a credentialist field. Product management has a prestige problem. The industry places enormous weight on what companies you worked at, which bootcamps you attended, which courses you completed. Yang thinks this is mostly nonsense. He hires for high-agency people who have built things and shown results - and says so explicitly, repeatedly, in a field where that view still carries some risk.
He spends half his time talking to customers. This sounds obvious but almost nobody does it. Yang has talked explicitly about the tension between building in public vs. building for your audience, and his solution is simple: ask them what they actually need. The 100+ AI tutorials in his archive didn't emerge from Yang's imagination - they emerged from the questions 148,000 people kept sending him.
He names failures without euphemism. Reddit Talk was discontinued. That's not a pivot or a learning exercise in the newsletter - it's a story about building something that didn't fit the platform's broader strategy. His three-year PM job search is not a "winding journey" - it was hard and nearly broke him. This directness is rarer in creator content than it should be, and it explains a lot of the loyalty his audience has built over time.
He sweats the small stuff. The default states, the edge cases, the copy. Yang's product philosophy is deeply anti-"ship it and iterate" in the sense that the iteration he cares about is at the detail level, not the feature level. Ship things that feel done. Fix the things nobody else notices. His readers have internalized this - it shows up in how his newsletter is written, in how his course materials are structured, in how he describes other people's products.
He builds in public with his family. Somewhere in his GitHub is a game he built with his 7-year-old. It's not a showcase project. It's just there, quietly demonstrating that he actually does the things he talks about - and that the definition of "builder" extends beyond the professional resume.
Yang has articulated the creator economy's core problem as a power law problem: the top 1% of creators capture 99% of the income. The same dynamics that made Twitch Raids valuable - helping small streamers get discovered, flattening some of that curve - animate what he's trying to do with Behind the Craft.
If practical AI literacy becomes table stakes for builders and creators in the next five years, then the people who had access to clear, actionable education early will have a meaningful advantage. Yang is trying to be that access point. Not with a course-first, community-first, conversion-funnel model - with a newsletter that is genuinely good enough that people recommend it to colleagues without being asked.
He's also watching the AI era reshape the PM role in real time from inside Roblox, and translating what he sees directly into the newsletter. The feedback loop between his day job and his media platform is tight. What he learns shipping creator tools in the morning shows up in his tutorials by Sunday.
The aspiration is straightforward: close the knowledge gap between the people building the AI era and the people trying to navigate it. Teach practitioners the way practitioners want to be taught - with specificity, without hype, and with enough humor that it doesn't feel like a chore to read.
A PM's job is to empower their team to build products that solve customer problems and make business impact.
- Peter Yang