Robots Are Coming! (No, They’re Not!)

Hello, and happy Sunday! We’re off on Monday for Columbus Day/Indigenous Peoples’ Day in the U.S. See you back here on Tuesday.—Kate Lee

Was this newsletter forwarded to you? Sign up to get it in your inbox.

Knowledge base

"How to Use Claude Code for Everyday Tasks—No Programming Required" by Katie Parrott/Source Code: That black window developers type code into can be your secret weapon, too. Claude Code—the terminal version of Claude—removes all the annoying limitations from the web app and gives you AI with the training wheels off. Katie Parrott shows how Every's team uses it to analyze massive datasets, build expense trackers, and answer customer questions without bothering engineers. Read this if you want to seriously level up your AI skills.

"Seeing Science Like a Language Model" by Dan Shipper/Chain of Thought: For centuries, we've been forcing psychology, economics, and social sciences into a straitjacket—trying to reduce human behavior to simple equations and universal laws. Dan Shipper argues that language models reveal a radically different way forward. Read this if you want to understand how AI isn't just a tool for doing science faster—it's ushering in a fundamental paradigm shift in how we pursue knowledge itself.

"Smuggled Intelligence: Why AI Progress Is Real and Most Jobs Are Safe" by Dan Shipper/Chain of Thought: Are robots really coming for our jobs? While GPT-5 Pro is solving math problems that stumped previous models and OpenAI's new benchmark shows AI matching human experts nearly half the time, there's a secret ingredient behind these achievements: humans. Dan reveals the “smuggled intelligence” that makes these AI feats possible. Read this to understand why the future of work isn't jobless, but transformed.

"Vibe Check: OpenAI DevDay 2025" by Dan Shipper and Alex Duffy/Vibe Check: Dan and Alex Duffy report from OpenAI's third annual DevDay, which they said felt like watching a magician explain their tricks. The company launched a new app store, AppsSDK, a workflow builder called AgentKit, and API access to its best models, including GPT-5 Pro and Sora 2. Read this if you want an insider's perspective on whether OpenAI is still pushing boundaries or just polishing what already exists.

"How Box Is Building an AI-first Company—Without Cutting Jobs" by Rhea Purohit/AI & I: Box CEO Aaron Levie reassures those still afraid AI will replace them: Agents are more like interns who never call in sick but occasionally make stuff up. In this episode of AI & I, Levie explains why jobs are more than just tasks, and how easier tasks mean more valuable work. Listen to, watch, or read about this if you want to understand how established companies can become “AI-first” without pink slips, and why every knowledge worker is becoming a manager of AI agents. 🎧 🖥 Listen on Spotify or Apple Podcasts, or watch on X or YouTube.

From Every Studio

The Sparkle website gets a new look

Sparkle is getting a visual upgrade. App general manager Yash Poojary and creative lead Lucas Crespo are redesigning the website to show how Sparkle connects file organization, search, and Mac commands into one system. The new homepage features an animated dock logo, files organizing themselves in real time, and upfront messaging about Sparkle's zero data retention policy—we don’t use your data or sell it to anyone else. Keep an eye out for the new look in the next few weeks.

Monologue launched teams in a weekend

When an Every superfan reached out wanting to know about team accounts for Monologue, general manager Naveen Naidu Mummana built the entire feature over a weekend. Monologue now supports team creation, billing (three seats minimum), and self-serve invites. Naveen is manually onboarding early teams and offering direct Slack support while gathering feedback. If you're interested in teams access, reach out to naveen@every.to.

Alignment

Sleep’s missing conductor. I’ve slept badly all week. My Oura ring showed my resting heart rate running a touch higher than normal and, yes, it’s probably because I went to the gym later than I usually would. (I also had Earl Grey tea at 6 p.m. a few days in a row because I’m British.)

But I don’t want to have to play amateur detective on my wearables’ graphs and numbers—especially not when I’m underslept. What I really want is a personal health agent that reads my day like a story and hands me concrete next steps like, “Ash, stop working out late,” or “Stop drinking tea after 2 p.m.”

That’s why I’m excited by the preliminary findings from Google’s research paper, "Anatomy of a Personal Health Agent." Its agent orchestrated three specialists to work in harmony: a data analyst to crunch your numbers from a wearable device, a medical expert to give those numbers clinical meaning, and a health coach to translate insight into simple next steps. Both end users and domain experts preferred Google's Personal Health Agent to either a single generalist bot or a parallel setup that mashes up answers from multiple bots without dynamically assigning “main” or “secondary” roles.

The findings suggest that AI agents, like humans, thrive with clear roles and hierarchies. What works about a family doctor coordinating your care with specialists works here, too. Even as we push the boundaries of artificial intelligence, the most effective systems seem to mirror existing organizational principles for human intelligence.

Which means I might soon have my long-wished for AI health coach—and be sleeping like a baby.—Ashwin Sharma

That’s all for this week! Be sure to follow Every on X at @every and on LinkedIn.

We build AI tools for readers like you. Write brilliantly with Spiral. Organize files automatically with Sparkle. Deliver yourself from email with Cora. Dictate effortlessly with Monologue.

We also do AI training, adoption, and innovation for companies. Work with us to bring AI into your organization.

Get paid for sharing Every with your friends. Join our referral program.