📝 TL;DR 📝 Runway has added Agent Skills to Runway Agent, its conversational marketing tool. Type / inside the Agent, pick a packaged workflow like "build an ad campaign" or "localize ads," and it executes the full multi-step process automatically. This sits on top of Agent 2.0, released in late June, and is aimed squarely at marketers scaling creative output. It is available now for paid Runway users. 🧠 Overview 🧠 Runway Agent launched as a conversational AI creative partner: describe your product or campaign, and it works out the angles, concepts, and assets likely to perform. Agent Skills is the next layer on top of that conversational foundation, and it addresses a familiar problem with any AI chat tool used for repeated business tasks: typing out the same detailed brief every single time is tedious, even when the AI understands you well. Skills solve that by packaging a specific, repeatable multi-step workflow behind a single slash command. Instead of describing what an ad campaign needs from scratch each time, you type /, select the skill, and the agent runs its established process automatically. This is a familiar pattern if you have used slash commands in Claude Code or other AI tools, now applied specifically to marketing and creative production. 📜 The Announcement 📜 Runway positions Agent 2.0, which launched in late June, as a tool that helps marketers "make more of what works": ads, videos, and full campaigns intended to drive revenue. According to Runway's own materials, the Agent can generate marketing briefs, produce campaign assets, and analyze performance data within a single ongoing workflow, functioning less like a single-purpose generation tool and more like an end-to-end creative partner. Agent Skills builds directly on that foundation. Rather than relying purely on conversational prompting for every task, users can now invoke pre-built, packaged workflows for specific job types, building an ad campaign, creating a commercial, and localizing ads across markets are the examples currently highlighted. Each skill runs its full multi-step process on a single command rather than requiring the user to manually direct each stage.

📝 TL;DR 📝 Meta has quietly soft-launched Pocket, an app that turns a text prompt into a playable, interactive AI-generated mini-experience called a "gizmo," no code, no game engine, nothing to learn. It comes from the team behind Gizmo, a vibe-coding app Meta acquihired earlier this year. The catch: it is not available in the US yet, and Meta has not made any formal announcement. This is a "watch where consumer AI creation is heading" story more than a tool you can try today. 🧠 Overview 🧠 Vibe coding, describing what you want in plain language and letting AI build it, has mostly lived in developer tools until now. Pocket is Meta's attempt to bring that exact experience to a completely non-technical, social audience. Instead of a code editor, you get a prompt box. Instead of a repository, you get a scrollable social feed of things other people made. The pitch is genuinely simple: type "turn a flower into a paintbrush so I can draw with it," and Pocket builds a playable version you can try immediately, tweak, and share. This launched with no press release and no announcement from Meta. It was spotted by reverse engineer Alessandro Paluzzi on July 2, and reporting from Appfigures traces the actual launch back to June 29 on both the App Store and Google Play. Meta has not responded to press requests for comment, which is consistent with a company running an early, low-key experiment rather than a flagship rollout. 📜 The Announcement 📜 Pocket describes itself as "a creative platform for making and sharing gizmos." A gizmo is Meta's term for an interactive, playable AI-generated experience built from a single natural language description. Meta's own example prompt is "make a drawing gizmo where the flower is the paintbrush," and the app generates a working, playable version from that description alone. Gizmos can respond to touch and phone tilt, play sound and music, and pull from the camera or photo library, giving them real interactivity rather than a static generated output.

📝 TL;DR 📝 Claude Fable 5 returned globally on July 1 after the US export controls that forced it offline on June 12 were lifted. It comes back with a more aggressive cybersecurity classifier, one that will flag a slightly higher share of harmless coding requests as a deliberate trade-off for blocking more genuine misuse. If a normal request gets flagged, use the feedback button rather than assuming something is wrong with your prompt. Paid plans get Fable 5 included up to 50% of weekly usage through July 7, then via usage credits. 🧠 Overview 🧠 The most powerful Claude model got pulled offline for nearly three weeks after a US government export control directive, and it is now back with real changes underneath the hood. Understanding what actually happened, and why the new version behaves slightly differently, is useful both practically (you might hit a refusal on something ordinary) and as a window into how AI safety actually gets tuned in response to real-world findings rather than theoretical ones. Anthropic's own explanation of what happened is unusually detailed and worth understanding directly, because it clarifies something that got muddled during the outage: the vulnerability behind the shutdown was not unique to Fable 5 at all. 📜 The Announcement 📜 As of July 1, Fable 5 is available globally across Claude.ai, the Claude Platform, Claude Code, and Claude Cowork. For Pro, Max, Team, and select Enterprise plans, it is included for up to 50% of weekly usage limits through July 7, after which access shifts to usage credits. Mythos 5, the less-restricted sibling model, has been restored only to a specific set of US organizations approved by the government on June 26, with broader Project Glasswing access still being negotiated. Anthropic's account of the underlying trigger is notable: Amazon researchers had found a way to prompt Fable 5 into identifying software vulnerabilities and, in one case, producing code demonstrating how to exploit one of them. But Anthropic's own testing found that less capable models, including Opus 4.8, GPT-5.5, and Kimi K2.7, could identify the exact same vulnerabilities.

📝 TL;DR 📝 Anthropic released Claude Sonnet 5 on June 30, a more agentic, near-Opus-performance model at a much lower price, now the default for Free and Pro plans. On the same day, the US Commerce Department lifted the export controls that forced Fable 5 and Mythos 5 offline on June 12. Anthropic says Fable 5 returns globally on Wednesday, July 1, with Mythos 5 access restoring more narrowly to vetted partners. It has been a genuinely eventful 48 hours for anyone using Claude. 🧠 Overview 🧠 Two significant Anthropic stories landed within a day of each other, and it is worth understanding them as connected rather than coincidental. First, the resolution of the Fable 5 and Mythos 5 shutdown, which has hung over the Claude ecosystem for nearly three weeks after a Commerce Department export control directive forced both models offline worldwide. Second, the launch of Claude Sonnet 5, a genuinely significant mid-tier model release that landed the same day the export controls were lifted, giving Anthropic a strong second announcement to pair with the restoration news. For anyone building on Claude or using it daily, this is the most consequential 48-hour stretch for the platform since the original shutdown. 📜 The Announcement 📜 Claude Sonnet 5: Anthropic released Sonnet 5 as, in its own words, "our most agentic Sonnet yet." It is now the default model for Free and Pro plans and is available to Max, Team, and Enterprise users, as well as through Claude Code and the Claude Platform. Pricing starts at an introductory $2 per million input tokens and $10 per million output tokens through August 31, then moves to standard pricing of $3 input and $15 output per million tokens, a meaningful discount to Opus 4.8's $4 input and $25 output pricing. On benchmarks, Sonnet 5 closes much of the gap with Opus 4.8 while remaining substantially cheaper. On one agentic coding benchmark, Sonnet 5 scored 63.2% compared to Opus 4.8's 69.2% and the previous Sonnet 4.6's 58.1%. On a knowledge-work benchmark, Sonnet 5 actually slightly outperformed Opus 4.8. Anthropic also reports improved safety behavior, lower rates of deception, sycophancy, and hallucination compared to Sonnet 4.6, along with cyber safeguards enabled by default.

📝 TL;DR 📝 Aside is a new YC-backed AI browser that signs into your existing accounts and performs real tasks inside them, emails, dashboards, internal tools, rather than just reading public web pages. It runs locally with encryption, lets you bring your own Claude or ChatGPT subscription instead of paying for a separate AI bill, and claims top scores on agentic browsing benchmarks. The capability is genuinely impressive. It also raises real questions about agent permissions that are worth understanding before using it on anything sensitive. 🧠 Overview 🧠 Most AI browser agents today work by reading what is publicly visible on a page or relying on pre-built integrations with specific services. Aside is positioning itself differently: it logs into your actual accounts the same way you would, using saved credentials, and operates inside them directly. That means it does not need a custom integration for every tool you use. If you can log into a website and do something there, Aside is designed to be able to do it too. This is a meaningfully different category of capability than reading and summarizing web content, and it comes with a correspondingly different risk profile. The privacy framing, local processing, encrypted storage, credentials never exposed to the AI model itself, is clearly built to address the obvious concern that comes with granting an AI agent that level of account access. 📜 The Announcement 📜 Aside is a desktop browser, currently available for macOS, built by a Y Combinator-backed startup. The pitch is that it works across logged-in websites and accounts directly rather than depending on integrations, handling tasks like managing communications, replying to messages, working with documents and spreadsheets on your computer, and operating inside dashboards and internal tools. On benchmarks, Aside reports ranking first across three agentic browsing evaluations: Online-Mind2Web, BU-Bench-V1, and Odyssey. The company's own published numbers show Aside scoring 99.0% on these combined evaluations, ahead of Browser Use at 97.7%, GPT-5.4 at 92.8%, Claude Opus 4.8 at 84.0%, and ChatGPT Atlas at 70.0%. These figures come directly from Aside's own marketing materials and have not yet been independently verified by a third party, which is worth keeping in mind before treating them as settled fact.