Apple's next AI move

PLUS: Google releases Gemini 3.1 Pro

Author

Barsee
23 Feb

Together with

Howdy, it’s Barsee again.

Happy Monday, AI family, and welcome back to AI Valley.

Today’s climb through the Valley reveals:

  • Google releases Gemini 3.1 Pro

  • Google releases AI product image generator and vibe-coding model

  • OpenAI's first AI device could be a smart speaker

  • Apple's next AI move

  • The AI companies are setting a new standard

  • Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

FLOW

Courtesy: Flow

Wispr Flow turns your voice into ready-to-send text in any app. Give coding agents 10x more context by speaking instead of typing. Works in Cursor, Claude, ChatGPT, and every IDE.

89% of messages sent with zero edits. Millions of users worldwide. Available on Mac, Windows, iPhone, and now Android.

Free and unlimited on Android during launch.

*This is sponsored

THROUGH THE VALLEY

Google has released Gemini 3.1 Pro, a new AI model designed for complex reasoning and multi-step tasks. Built on the Gemini 3 architecture, it scored 77.1% on the ARC-AGI-2 benchmark, more than doubling the previous Gemini 3 Pro result and beating models from Anthropic and OpenAI. Developers can adjust how much the model “thinks,” choosing faster responses or deeper analysis depending on the task. Gemini 3.1 Pro is available through the Gemini API, Vertex AI, and Google’s new Antigravity IDE, with early demos showing it can build animations, simulations, and long-running agent workflows from simple prompts.

Google Labs has launched Photoshoot, a new feature inside its free Pomelli tool that helps small businesses create clean product photos without a studio. Using a brand profile and Google’s Nano Banana image model, the tool can turn simple phone photos into polished product images, with options to change backgrounds and match a brand’s style. Google is also working on a major upgrade to AI Studio’s Build feature, which aims to support full app development with real user sign-in, Firebase authentication, third-party API connections, and better support for frameworks like Next.js, making it more practical for building real products.

OpenAI is preparing its first consumer hardware product with designer Jony Ive, and it is expected to be a smart speaker priced between $200 and $300, according to The Information. The device will reportedly include a built-in camera and facial recognition to handle tasks like purchases. A team of more than 200 people is working on the project, with an early 2027 launch target. The group formed after OpenAI acquired Ive’s startup for $6.5 billion, bringing in several former Apple leaders. OpenAI is also exploring smart glasses and other AI devices, but the speaker will be its first real test in consumer hardware.

Anthropic has launched Claude Code Security, a new feature inside Claude Code now in limited research preview. The tool helps security teams and open-source maintainers find software bugs by analyzing code in context, rather than relying only on fixed rules like traditional scanners. It can catch issues such as access control mistakes and business logic flaws that are often missed. Built on Claude Opus 4.6, the system assigns severity and confidence scores and suggests fixes, which developers still need to review. The feature is available to Team and Enterprise users, with priority access for open-source projects, as Anthropic prepares for more AI-driven security threats.

Apple’s next push into AI-powered wearables will center on “Visual Intelligence,” a system designed to understand what you see and use that context to help you in real time. CEO Tim Cook has hinted that future devices will rely on on-device visual AI to interpret the surrounding environment and take useful actions. Apple is developing its own vision models to power a new lineup of products, including advanced AirPods, smart glasses, and even a pendant-style device. The goal is hands-free assistance that works in everyday settings, without needing to pull out a phone. Apple is expected to unveil its first products of the year in March.

TRENDING TOOLS

  • MongoDB for Startups gives you what you need to build, scale, and succeed when getting started. Learn more *

  • Replit Animation > Vibecode your next viral video in minutes, powered by Gemini 3.1 Pro

  • Grok Build > xAI’s browser-based coding workspace now runs on Grok 4.20’s multi-agent system, letting four AI agents collaborate on your code at the same time

  • AI Doctor > AI that knows more about your body than any human ever could

  • Rork Max > AI that one-shots almost any app for iPhone

  • Interpreter > Al-native workspace for editing Word docs, automating Excel, and filling PDFs

  • Manus AI > Now built into Meta Ads Manager, it automates reports, audience research, and campaign analysis directly inside your ad workflow

(*) signifies sponsored product

SIGNALS

The AI companies setting the standard

A small group of AI-native startups are becoming the benchmarks for how fast modern software companies can grow. What matters most is not their industry, but how they build, ship, and use AI inside their own products.

  • Became the fastest SaaS company to reach $500M in ARR.

  • Ships from a single codebase every 2 to 4 weeks, which keeps things simple and fast.

  • Is moving from being just an AI wrapper to a full platform.

  • Its in-house models now write a large share of production code.

  • Background agents let one engineer manage several AI coding agents at the same time, turning each senior engineer into a team lead for AI workers.

  • Reached $200M ARR in under a year with a team of around 30 people.

  • Built for non-technical founders who want to go from idea to product fast.

  • With one prompt, users can generate a full-stack app using React and Supabase, including login and database logic.

  • Turns MVP building into a near-instant process instead of a long technical project.

  • Hit $100M ARR in just 8 months, making it one of the fastest-growing vibe coding platforms so far.

  • Operates across San Francisco and Bengaluru.

  • Uses multiple specialized AI agents instead of one general model.

  • Different agents handle architecture, backend, and security, all working together in one workspace.

  • This gives teams a more structured and reliable way to build software with AI.

  • Reached close to $200M ARR in about three years by focusing only on legal AI.

  • Shows how strong vertical AI can be when it is built deeply for one profession.

  • Its Memory system adapts to how each law firm writes and thinks.

  • This turns Harvey into firm-specific software, not just a generic legal assistant, which makes it much harder for general models to replace.

WHAT I'M CONSUMING

INDUSTRY MOVES

  • Epoch AI reports Anthropic is growing 10x per year and could overtake OpenAI in revenue by mid-2026

  • OpenAI expects revenue to hit $30b this year and $62b next year, driven mainly by ChatGPT

  • OpenAI raises revenue outlook but predicts $111b more cash burn through 2030

  • ByteDance is expanding its AI team in the US, hiring nearly 100 people for its seed division with labs in the US, Singapore, and China

THE VALLEY GEMS

What’s trending on social today:

THAT’S ALL FOR TODAY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.