First Look

Apple is bringing its Apple Intelligence features to Chinese users via an iOS 27.3 beta launching 4 November, with a full release in December, according to The Verge.

The rollout covers notification summaries, real-time translation, writing tools, visual intelligence, and the ability to remove objects from photos. Apple has partnered with Alibaba to integrate the Qwen chatbot into some features — comparable to the ChatGPT integration available elsewhere since 2024.

Notably absent: the upgraded Siri. Apple trained a separate LLM (large language model — the AI system that generates text) specifically for the Chinese market, a requirement driven by local regulatory constraints. The beta is available to developers next week.

Who it's for: iPhone users in mainland China who have been waiting while every other market got these features first. For everyone else, this is mostly a market access story.

Honest read: The feature set is meaningful for Chinese users who have been locked out of Apple Intelligence entirely, but the absence of the new Siri makes this a partial launch rather than a full one — worth watching to see whether that gap closes or stays permanent.

Google Gemini — Google's AI assistant — can now generate, preview, and edit files directly within the chat interface, according to The Verge. Supported formats include Google Docs, Google Sheets, slide decks, PDFs, Markdown, plain text, and code files.

Previously, Gemini could create a file and let you download it — but editing meant opening a separate app. Now you can highlight a section of text and prompt Gemini to rewrite it, or select a table and ask it to convert it to a chart, all without leaving the conversation.

PDF editing works on both web and mobile today. Other file formats are web-only for now. The update is available to all Gemini users immediately at no extra cost. Anthropic's Claude launched similar in-chat document editing earlier this year.

Honest read: Worth trying if you draft documents regularly in Google Workspace — the ability to edit without context-switching is a real time saver, not just a demo feature.

Apple's iWork suite — its free productivity apps for writing, presentations, and spreadsheets — is getting a built-in AI assistant, according to The Verge.

The assistant is built on Apple's own models plus OpenAI's, and it understands the document you have open. In Pages it can draft and restructure text. In Numbers it suggests chart types and builds tables. In Keynote it can animate and redesign slides from a prompt.

It is rolling out in beta to devices running iOS 27, iPadOS 27, or macOS 27 that support Apple Intelligence — so not everyone has it today. The apps also now support the Model Context Protocol (a standard that lets third-party AI agents plug into other software) on macOS 27, meaning outside tools can also interact with your documents.

It is free, bundled with the apps that are themselves free.

Honest read: Worth trying if you are already in the Apple ecosystem and find yourself doing repetitive formatting or drafting work — but this is a beta, and Apple's description of its own capabilities is all the evidence available so far. No independent evaluation yet.

OpenAI has made its Decisions API generally available, according to the product announcement. It is a programming interface — a way for developers to connect their software to OpenAI's capabilities — built on GPT-6 Luna and designed for a specific job: fast, repeated yes/no and classification questions at volume. Think routing a customer support ticket, tagging a financial transaction, or choosing which response to serve next.

The API can handle up to 400 questions per request and integrates with OpenAI's live voice product, GPT-Live. A new prompt caching feature reduces costs when the same context is reused. Inference residency in the US and EU is included.

OpenAI claims it is up to 10x faster than querying GPT-6 Luna through its standard interface — that figure comes from OpenAI's own testing, not an independent evaluation. Pricing details are in the availability guide.

Honest read: Potentially useful if you are a developer running the same classification question thousands of times a day. If you are not building an application, this does not change anything for you.

1Password is a password manager — an app that stores your login credentials and fills them in automatically. Its new feature, 1Password for Agents, lets AI agents (automated software programs that complete tasks on your behalf) sign into websites and services without the agent ever seeing your actual password.

Here is how it works: when an AI agent hits a login page, it uses a unique digital key to identify itself to 1Password. The app sends you a notification asking for approval, you say yes, and 1Password fills in the credentials directly. The agent gets in, completes its task, and never handles the password itself. It works similarly to two-factor authentication — you confirm the action on your own device.

It currently supports agents including Meta Muse, ChatGPT Work and Dots, Claude Duet, and Grok Bot, with more partners announced.

There is no mention of a free tier in the announcement, so pricing follows 1Password's standard subscription.

Honest read: If you are already using AI agents for real tasks — booking travel, managing email, filing expenses — this is a meaningful safety improvement over pasting credentials into a chat window. Worth enabling as soon as your agent supports it.

Adobe Lightroom mobile — a photo editing app for iOS, iPadOS, and Android — can now open closed eyes in photos using generative AI (AI that creates new content rather than just adjusting existing pixels). When the app spots a blink in a group shot, it places a one-tap shortcut over the image. Tap it, and the eyes are replaced with a version that matches the lighting, pose, and — using other photos of that person in your library — the correct eye colour, size, and shape.

This is a meaningful step up from older tools like Photoshop Elements, which sourced replacement eyes from a separate reference photo with inconsistent results.

The feature is available now on the free version of Lightroom mobile, with two free uses before a Creative Cloud subscription is required. Similar fixes exist in full Photoshop, but they take more steps and are not free.

More details at The Verge.

Honest read: Worth trying if group photos are part of your life — two free attempts is enough to know whether the quality justifies a subscription.

Bluesky — the decentralised social platform — launched a tool called Attie earlier this year to let users build custom content feeds using AI. According to The Verge, Attie has now expanded to let anyone create small shareable apps: polls, games, dashboards, and more, each hosted on its own attie.site link.

You type a prompt, Attie builds it. Examples already live on the site include a Flappy Bird clone, a drum machine, and a quiz. There is no coding required.

This is aimed at Bluesky's existing community — people who want to add something interactive to their corner of the platform, not developers building serious software. It is free to use. The output, based on early hands-on reports, is lightweight; functional novelties rather than polished products.

Honest read: Worth a look if you are already on Bluesky and curious — but the output quality sits closer to a fun experiment than a real mini-app, so expectations should match accordingly.

Meta has published Muse Spark Open, according to Hugging Face — an open-weights model (meaning you can download and run it, though the training data is not public) built specifically for coding tasks. It can read a codebase, write and debug code, process images, audio, and video, and call external tools, all within one model.

It is primarily aimed at developers, not general users. The full version requires hardware most individuals won't have access to — think multiple server-grade GPUs and roughly 1.5 terabytes of storage. This is a tool for teams and researchers, not a laptop download.

What it does that others don't: it handles multiple input types alongside code in a single model, and it plans and self-corrects across long tasks rather than answering in isolated turns.

It is free to download under Meta's licence. Verify licence terms before any commercial use.

Honest read: Worth watching if you work in software development or AI research — genuinely capable on paper, though independent evaluations haven't landed yet, so treat Meta's own capability claims accordingly.

Gemini 4 Argon is Google's first major frontier model (a top-tier AI system competing at the cutting edge of the field) since Gemini 3 in late 2025, according to The Verge.

Google says Argon is built for complex, multi-step tasks — coding, finance, legal work, science, and cybersecurity. The company claims it outperforms rival models including GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 across a range of benchmarks, though those results come from Google's own testing, not independent evaluation.

It is live today for Pro and Ultra subscribers in the Gemini app and in AI Mode in Search. Ultra subscribers also get access via Antigravity, Google's agentic development platform. Free and Plus users have no confirmed timeline.

Pricing starts at $2 per million input tokens and $10 per million output tokens — roughly one-fifth the cost of GPT-6 Astra, according to Google. That introductory rate doubles at the end of November.

Honest read: Worth trying if you are already a Pro or Ultra subscriber — Google has something competitive here, though independent benchmarks will tell the real story.

Anthropic has launched OSS Scanner, a free service that runs automated security scans on open-source software projects. According to The Verge, projects that opt in receive periodic vulnerability reports generated by Anthropic's strongest models, including one called Claude Mythos.

The service is aimed at open-source maintainers who lack the resources for regular security audits. There is no cost. The trade-off is that reports are entirely model-generated with no human review, which means false positives are a real possibility. Anthropic says the speed and frequency of scanning is the upside; accuracy is the honest caveat.

AI-assisted bug hunting has found genuine flaws before — a recent example being a vulnerability affecting nearly every Linux distribution. But some open-source communities are already struggling to manage floods of low-quality AI-generated bug reports, which adds overhead rather than reducing it.

This is opt-in, free, and genuinely useful for under-resourced projects. The absence of human review is a real limitation, not a footnote.

Honest read: Worth trying for small open-source projects that currently have no security scanning at all — go in knowing the reports need human judgement before acting on them.