Week in Product #488 🚀
AI prototyping, OpenAI's cybersecurity platform, Data centers in space, AI Agents on the App Store, Building a personal OS, Agentic engineering playbook, Codex mobile & more
Hi folks 👋
Welcome to a new Week in Product!
🎰 The week in figures
$70B: Cerebras, the AI chep leader, had the biggest IPO of 2026 with stock surging nearly 70%, sending the company’s market cap to almost $70B
$750M: Ramp is in talks to raise another $750M at a pre-money valuation of +$40B
$650M: Recursive Super AI, a startup by ex-leaders from Meta AI, Google DeepMind, OpenAI, raised $650M at $4.65B valuation targeting self-improving AI
4K: Cisco cuts 4,000 jobs despite record revenue: Networking giant slashing 5% of workforce to fund AI and cybersecurity investments
94%: Baidu's new AI model cost 94% less to build and outperforms chinese competitors
34.4%: Ramp's May 2026 AI Index shows Anthropic at 34.4% of business adoption, crossing OpenAI (now at 32.3%) for the first time ever
5%: LinkedIn planning to lay off 5% of staff, around 875 people, blaming the need to “reinvent” how it works and building more “agile teams focused on [its] highest priorities”
📰 What’s going on
OpenAI shakes up its product team. Greg Brockman will now handle both the company’s AI infrastructure and its product teams, which are being consolidated into a single org. Now, one group will manage ChatGPT, Codex, and developer-facing APIs
OpenAI is preparing to sue Apple after the ChatGPT integration, announced at WWDC in June 2024, failed to deliver the subscriber growth and product prominence the company expected. OpenAI has enlisted an outside law firm to work through its options, which could include sending Apple a formal breach-of-contract notice
OpenAI has unveiled Daybreak, a cybersecurity platform designed to automate vulnerability detection, patch validation, and secure software development for enterprises and governments. The platform directly competes with Anthropic's Claude Mythos and combines OpenAI's LLMs with Codex's agentic capabilities
Anthropic limits Claude usage as costs spike: Paid subscribers now face separate credit meters for agent tools after companies like ServiceNow and Uber burned through annual AI budgets
Anthropic launches Claude for Small Business: Package includes 15 ready-to-run workflows and integrations with QuickBooks, PayPal, HubSpot, Canva, and Docusign to automate payroll, invoicing, and campaign management for 44% of U.S. GDP that relies on small businesses
AWS launches Claude Platform on AWS. Direct access to Anthropic's native platform through existing AWS accounts with IAM authentication, CloudTrail logging, and full Messages API support
Google preps "Gemini Spark" agent for enterprise: Beta code reveals autonomous agent with cross-app access, scheduled tasks, and inbox management; launching ahead of I/O 2026
Google DeepMind introduced "AI Pointer," a concept that turns the traditional mouse cursor into a context-aware AI collaborator that understands what's on your screen and can act on it, starting with Chrome
Google and SpaceX in talks to put data centers into orbit, part of Project Suncatcher with prototype satellites by 2027
Apple is building AI agent support into the App Store, with a WWDC announcement expected; it would mean AI agents (not just apps) could eventually be found, downloaded, and run directly from your iPhone
📚 Good reads
Daily Active Agents. Baidu’s CEO proposed “Daily Active Agents” (DAA) as the defining metric for the AI era, the equivalent of Daily Active Users for the agent world, and predicted global DAA could eventually surpass 10B
Why bolt.new currently leads AI prototyping tools for PMs. Aakash Gupta runs a detailed bakeoff of Bolt, Lovable, Replit, and v0, finding Bolt the most production-ready and brand-faithful. He also distills some workflows (including baseline design systems, divergent prototyping, PRD‑first builds) that improve results in any tool. The piece closes with practical “which tool + workflow when”, and common mistakes
IC work is the new career flex. Elena Verna argues that, thanks to AI, experienced “high-impact ICs” can now ship end‑to‑end projects solo, having department-level impact without managing people. She explains how AI gives you “average intelligence” across adjacent crafts so you can prototype, validate, and iterate way faster than old coordination-heavy orgs. A proof: the app I just launched (see below)
🎧 Good listen/watch
[Podcast] Build a Claude Code personal OS in 40 minutes. Moritz Kremb shows how he runs email, content creation, and even grocery shopping through a Claude Code–based “personal OS.” He layers it into folders with memory, connected tools (APIs/MCPs/CLIs), reusable “skills,” and proactive routines. The episode also digs into OpenClaw vs Claude Code tradeoffs, reliability issues that pushed him to Claude, and concrete examples like automated grocery runs and a full content pipeline from idea to published video
🧑💻 Worth learning
Agentic engineering playbook. This piece explains how agent skills can encode your company’s unique processes, systems, and data into reusable “skills” that coding agents can reliably invoke. Over time, the org’s job may become building, sharing, and maintaining a large skill portfolio that drives serious productivity gains and more autonomous AI work
🔧 Products to try
Codex in ChatGPT mobile helps you steer coding tasks from your phone, review changes, approve commands, and keep long-running work moving while your laptop does the heavy lifting;: included across all ChatGPT plans
Claude Code Agent View gives you one dashboard for managing parallel Claude Code sessions and longer-running coding work, available on paid Claude plans
👨💻 My work
Henry, the baby tracker. This is not for everyone, but I hope parents will like it. Turns out sleep-deprived nights looking after a newborn can be productive. I didn’t like any of the baby tracker options on the market. They are overly complicated, trying to monetize too hard, or plain ugly. So I built one (iOS app) with just the things I need, caring for a simple and smooth UX to track my baby KPIs (read Key Poop/Pee Indicators) with one hand in the middle of the night. Includes a voice feature, so you can simply tap your lock or home screen and say what you want to log. It’s on TestFlight, and you can sign up here if you want to try it. And here is the story of how and why I built it. I hope other new parents find it useful!
That’s a wrap for this week.
Feel free to drop your comments/questions/feedback. Would love to hear what you’d like to see more about in WiP, so I can make it better for you.
Have a great weekend!
