Every Inc.

More people are building software—but not in the way you might expect. They're not engineers. They're writers, researchers, operators. People with deep context, persistent friction, and AI tools that finally speak their language. In this week’s Working Overtime, Katie Parrott reflects on learning to build with tools like Cursor and Lovable—despite having no coding background—and what vibe coding unlocks when you start from the problem, not the code. Inside: - Why the “technical vs. non-technical” divide is breaking down - How AI makes systems thinking more accessible - What a more expansive definition of building might look like Read the essay here: https://bit.ly/3YW8ix6

The biggest consumer success in AI so far? A single-player text box. But that’s just the starting point. In this week’s AI & I, Benchmark’s Sarah Tavel joins Dan Shipper to help chart what comes next. We've seen this pattern before—at Google, Facebook, Pinterest. The first breakthrough is usually technical. But the real winners will build products that feel right: intuitive, social, status-generating, fun. Also in the episode: - Why ChatGPT feels like Google in 1999 - What Custom GPTs get almost right - The $10B opportunity in multiplayer AI - And why the next great product might come from a power user, not a PhD If you’re building in AI—or just watching closely—this is one of those conversations that helps you see the path a little more clearly. 🎧 Full episode here: https://bit.ly/4lZGCBy

When new technologies emerge, we don’t invent new art forms right away. We film stage plays. We make “animated photographs.” It takes time (and tinkering) before a medium shows us what it really wants to be. Generative AI is still in its “stage-play era.” Most models crank out essays, briefs, or marketing copy that look familiar. But the real leap comes when we treat AI as a fresh canvas for completely new forms of storytelling. That’s exactly what novelist Eliot Peper is exploring at Portola, home of the Tolans—Pixar-cute alien companions (800 k downloads and counting) that chat, remember, and evolve right alongside you. In this week’s Thesis, Eliot lays out three field-tested principles for building stories that think for themselves: - Grow lore like a city. Instead of a rigid “canon bible,” let facts, festivals, and back-stories sprout organically from user questions, the way real neighborhoods fill in over time. - Seed situations, skip scripts. Give your AI character a concrete dilemma (“my cousin’s jealous I got matched with a human friend”) and let the back-and-forth co-write the plot beats. - Recycle memories into new moments. Prompt the model to mine past chats for callbacks, remix details, and push the narrative forward, so every conversation feels both personal and serialized. Read the full essay here → https://bit.ly/4jybUOh

Social platforms were built on the assumption that what we click is who we are. But as Dan Shipper argues in his latest Chain of Thought, AI systems flip that script, rewarding what we say we want, not just what keeps us scrolling. That subtle shift opens up a bigger question: What kind of internet do we build when the algorithm actually believes us? 📖 A thoughtful read on AI, alignment, and the future of recommendation systems to kick off your weekend: https://bit.ly/4cTy6jk

Most people don’t organize their files. They just… accumulate them. Downloads folder? Chaos. Desktop? A minefield. That’s what Sparkle is built for. It’s a Mac app that automatically organizes your files into a clean, structured system—without you having to think about it. Today, we’re launching Sparkle v2: rebuilt from the ground up for speed, clarity, and joy. It’s already sorted over 10 million files for users who’d rather do their work than manage it. 📁 Smarter logic ⚡️ Faster onboarding ✨ A new UI that feels Mac-native Whether you’re a founder, freelancer, or just someone with a cluttered desktop, you’ll feel the difference. 🛠 Try Sparkle v2 → https://bit.ly/3GkCFal P.S. If you’re curious how we rebuilt the whole thing in two weeks using AI (and caffeine), Spiral GM Yash Poojary tells the full story for Every. Link in the comments.

What does it mean to be a techno-optimist in 2025? Kevin Kelly—founding editor of WIRED and one of the most enduring voices in future-focused tech—joined AI & I for a wide-ranging conversation with Dan Shipper. They talked about the tools we build, the ways they shape us, and how to stay curious across decades of change. 🧠 How AI reshapes creativity, ambition, and the editing process ✍️ The shift from pioneer to builder—and why structure isn’t the enemy 📚 Annie Dillard, Burning Man, and the quiet power of attention 🛠️ Using ChatGPT to write ten novels for the joy of it This episode is thoughtful, expansive, and full of grounded wisdom for anyone thinking about the future—especially those trying to build it. 🎥 Full episode here: https://bit.ly/4juocr0

This week brought a cascade of releases from OpenAI: o3, GPT‑4.1, and o4‑mini—each with distinct capabilities. Meanwhile, Anthropic launched a new research tool and Dan Shipper talked to an AI alien that gives relationship advice. It’s a lot. So we broke it down in this week’s Context Window: 🔹 What each new OpenAI model actually does 🔹 Why o3 might be the most “agentic” model yet 🔹 How to choose the right one for your workflow 🔹 Why Anthropic’s release matters (even if it’s not flashy) 🔹 And what the rise of AI companions says about the future of software 📬 Read the full edition here: https://bit.ly/4jJwo6h

This is the perfect time to share an exciting personal news: I recently joined Every as an Engineer at Cora. In fact, this is the first big feature that I shipped! For those who know me as a "community guy", this marks a stark U-turn in my career. A scary career pivot. Last time i got paid to code was 6 yrs ago at an internship in college. Here's my wild career story: • After college, I decided to ditch my programming degree because that internship made me loose faith in all coding jobs. They all felt like being a cog in the wheel. But I wanted to do something more entrepreneurial. • So instead, I became a Community Manager at a small edtech startup. I dove into the wonderful (and very non-technical) world of community building. I ran communities, hosted events, and basically lived in the DMs. • Then 2 yrs ago, with the release of ChatGPT something clicked. I started building a tool I wished I had as a Community Manager. Even dragged my brother Piyush into it. • We taught ourselves Ruby on Rails - the most entrepreneur-friendly web development framework. Together, we built Curated Connections — our very own SaaS app. • And somehow, it became a competitor to startups backed by millions in VC funding. Not bad for two guys who were learning programming using ChatGPT. Over the last 2 years, we went from copying and pasting code from GPT 3.5 in ChatGPT to instructing agents in Cursor. • All this while, I’d been quietly fanboying over Every—a media company that gets what it means to blend AI, content, and software. They have created the most valuable AI bundle on the Internet today. • So last month, when they posted a job for a Rails dev, I had to apply – even though I wasn't "job searching". Things moved quickly. And here I am, working with the amazing Kieran Klaassen - creator of Cora - a new kind of email client built for this AI Age. I am insanely excited to be working with the team that's at the forefront of exploring what it means build media and software with the best of what AI has to offer. It's thrilling, scary and energizing - at the same time!

OpenAI just released a new model—and it changes what AI can do for your work. The model is called o3, and it’s now available inside ChatGPT. It’s faster, more capable, and—most importantly—it can follow through on multi-step tasks without hand-holding. In this review, Every CEO Dan Shipper puts o3 to the test and shows what it enables real users to do: ✅ Summarize hours of meeting transcripts—and surface leadership patterns ✅ Research guests for a podcast or project, pulling from web, social, and archives ✅ Build a custom learning course that pings you daily with bite-sized lessons ✅ Analyze your company’s org chart and predict what kinds of products you’re structured to ship ✅ Help you work through books, code, even blurry photos—without needing perfect prompts If you’re exploring what AI could actually do for your job, this is a great place to start. Link in the comments.

Ease makes people uneasy. AI exposes that. Why? Because we’ve spent centuries treating visible effort as proof of value. And now, tools that make work look easy are forcing us to question what actually counts as work. Today in Working Overtime, Katie Parrott explores what happens when writers gets dropped—not for bad work, but because an AI detection tool flagged their human-written article as machine-generated. The real issue, she argues? The work didn’t look hard enough. This piece traces our obsession with performative effort—from Puritan labor ethics to Slack green dots—and asks what it might take to redefine productivity in an AI-assisted world. 🧠 Rethink what work is. 🔍 Look past the performance. ⚙️ Make ease a feature, not a red flag. Read the full piece here: https://bit.ly/44pQEW0