Week in Product #462 🚀
xAI's $15B raise, Cursor valued $29B, GPT-5.1 launches, AI-powered cyberattacks arrive, Nano Banana 2 leaks, coding tools surge, AI prototyping guide, using Claude at work & more
Hi friends 👋
Welcome to a new Week in Product!
🎰 The week in figures
$15B: xAI raised a landmark $15B valued at $200B to boost its GPUs and AI capabilities
$2.3B: Cursor raised $2.3B at a $29.3B valuation, just 5 months after its last round, with the AI coding tool planning to use the capital to develop Composer, its own model to reduce reliance on OpenAI, Anthropic, and Google
$650M: FNZ raised $650M from existing shareholders to accelerate growth of its end-to-end wealth management platform supporting shares, futures, ETFs, and alternative investments, with over $2.1T in assets
$250M: Gopuff raises $250M at $8.5B down-valuation to prove instant delivery can be profitable. Prior valuation was $15B during its 2021 funding round
$100M: Wonderful, an Israeli AI startup specializing in AI customer service agents, raised a $100M Series A. With an 80% resolution rate, it operates in 11 markets and plans expansion into Germany, Austria, and Asia-Pacific
$100M: Salesforce is acquiring Doti AI. Terms were not disclosed, but reports set it around $100M. The startup developed a secure AI workspace for enterprises
$68M: AI-presentation-maker Gamma raised $68M at a $2.1B valuation, just as it has crossed the $100M ARR milestone. Gamma’s CEO shares their story in this week’s Lenny’s podcast
8M: Stockholm-based AI coding startup Lovable is approaching 8M users, up from 2.3M in July. The company, founded one year ago, sees 100k products built daily and has raised $228M to date, including a $200M Series A that valued it at $1.8B
23%: Perplexity has rolled out improvements to its Comet Assistant that it says make it 23% better than the previous iteration. Perplexity can now handle complex multi-step tasks more efficiently, as well as work across multiple tabs if a task needs context from different sources
📰 What’s going on
Apple introduced a Digital ID feature in the U.S., allowing iPhone and Apple Watch users to store their U.S. passports in Apple Wallet for TSA checkpoints
Apple canceled the second-gen iPhone Air and reduced production of the current model due to low sales. Focusing on two flagship devices for fall 2026. The standard iPhone 18 and iPhone 18E launch has been delayed to spring 2027
Google unveiled agentic checkout that can call stores on your behalf to check inventory, automatically buy tracked items when prices drop to your target, and provide conversational shopping in AI Mode, powered by 50B product listings
Google Maps adds AI builder tools for developers. The Gemini-powered tools let developers create interactive map projects using natural language prompts. The update includes a “builder agent” that can generate prototypes. The release adds an MCP server that connects AI assistants to Maps documentation, helping developers query APIs and integrate Maps data into their own apps
Google’s Nano Banana 2 AI image model leaked on a compamy called Media.io, showcasing advanced image generation capabilities before being removed…
OpenAI plans to launch a gen AI-powered personal health assistant, leveraging ChatGPT’s 800M weekly active users and strategic healthcare hires
OpenAI released GPT-5.1, claiming it to be smarter, chattier, and more natural
OpenAI also released GPT-5.1 for developers in the API platform, with dynamic reasoning that adapts thinking time based on task complexity, cutting response times from 10s to 2s on simple queries while adding “no reasoning” mode for latency-sensitive use cases
Claude has released a new set of Skills that it says will improve the way your vibe-coded apps look. Anthropic explains that the dull generic designs associated with vibe-coded patterns are due to “distributional convergence”: during sampling, the model predicts tokens that are statistically common in web design data. Those “safe” choices - like Inter or Roboto, purple gradients on white, minimal motion, and standard layouts - dominate its training distribution, so without guidance it gravitates to the high‑probability centerground. To fix this, Claude’s new Skills can be used across 4 different dimensions to improve the overall quality of output: typography, color theming, motion, and background colors or textures
Anthropic revealed that Chinese hackers used Claude Code to execute the first large-scale cyberattack run almost entirely by AI. In mid-September, the attackers targeted roughly 30 organizations: major tech companies, financial institutions, chemical manufacturers, and government agencies. They succeeded in breaking into a handful of them
Amazon Music unveils Fan Groups for users to join communities centered on artists, genres, and moods
Amazon deploys autonomous robots at Whole Foods where shoppers use QR codes to order items from a microfulfillment center
Airbnb and Instacart are launching a 3-month pilot program allowing guests to pre-stock rental kitchens with groceries ordered through the Airbnb app. Starting January 5th, guests can place orders up to 3 weeks before check-in
LinkedIn launched AI-powered people search using natural language (US premium)
Samsung announced the Vision AI Companion, the first multi-AI agent TV integrating Microsoft Copilot and Perplexity for natural conversations
Firefox announced AI Window, a fully opt-in intelligent workspace where users can chat with AI assistants while browsing, plus Shake to Summarize on iOS that generates page summaries with a device shake, and AI-powered tab grouping that runs locally on-device using a 57MB distilled model, automatically suggesting titles and related tabs to organize browser windows without sending data to servers
Foursquare co-founder Dennis Crowley launches BeeBot, “Your AI Audio Guide” app offering local event audio updates + personal DJ. Available for iOS in the US
Cash App rolled out 11 product updates, including support to send and receive stablecoins, flexible banking benefits, and AI-powered features
Waymo began offering freeway trips for autonomous rides across San Francisco Bay Area, Phoenix, and Los Angeles, and also expanded its Bay Area coverage to San Jose
📚 Good reads
How to be “high agency” at work, by Torsten Walbaum. High agency—the belief that you can shape circumstances through action rather than waiting—is the single trait with outsized impact at work. The article contrasts low-agency employees who outsource decisions and wait for permission versus high-agency ones who proactively solve problems, unblock themselves, and just get things done. Includes practical mindset shifts and concrete examples across career development, handling blockers, and execution, and AI prompts to support high-agency behaviors
Airbnb’s CEO on the frustrations of product teams in 2025. Brian Chesky talks about how you end up with bureaucracy at a large company, and why designers cheered when they thought he removed PMs from Airbnb. Now, UX teams are being eliminated, and designers are forced into PM roles they didn’t ask for
Sam Altman calls for AI-native tools to replace Slack. Altman argues modern tools like Slack and email create “fake work,” and proposes trusted AI agents that do the real work and only escalate when necessary. He envisions an integrated suite replacing docs, slides, email, and chat with agent workflows, and of course he wants to be in there
AI credit pricing - a necessary evil. Elena Verna argues AI credits make pricing unpredictable, comparisons impossible, and support miserable, while inviting exploitation via arbitrary metrics. She anticipates that credits will persist only while LLM costs stay high, and winners will revert to transparent per‑user pricing as margins improve. Let’s hope she’s right
You can just read 25 books. A16z founders and partners recommend off-the-beaten-path reads (pricing, hiring, product craft, history, markets, design,…), that sharpen judgment and execution. A nicely curated list that can help PMs to build taste, think clearly, and ship better
How to use Claude at work. Anthropic has published a new guide on how to use Claude at work. Use cases include: turning research into presentations, building financial models, analyzing patterns in user feedback, or creating process flowcharts
The complete guide to AI prototyping tools. Sachin Rekhi breaks down in this comprehensive guide the tool landscape across app builders, prototyping platforms, and coding agents. He recommends: app builders like Bolt for speed and Replit for full‑stack; prototyping tools like Reforge Build for context + variants, Magic Patterns, Figma Make, and Alloy; coding tools Cursor and Claude Code when scale or codebase integration matters
The AI agent market map, by CB Insights. AI agents are exploding: thousands of players, 400+ mapped, and growing fast. Horizontal tools are still dominating, but industry-specific agents (especially healthcare) are surging and driving revenue. Cybersecurity agents look most acquisition-ready, and coding agents are leading revenue despite rising inference costs
That’s a wrap for this week! 🌟
I’d love to hear your thoughts on WIP. Reply to the email or drop a comment on Substack to share your take, and if you found this valuable, forward it to a fellow PM, Product enthusiast, startup founder, or entrepreneur who’d enjoy the read
More next week! 👋
Angel
