In partnership with
Hey folks,
Anthropic dropped a nice surprise this week: double usage limits for everyone. Free, Pro, Max, Team, doesn't matter. For the next two weeks, you get 2x your normal limits outside peak hours. No codes to enter, nothing to click. It just works. The bigger news though is Claude's 1M-token context window is now live for everyone. You can now upload 600 images or PDFs or drop an entire codebase into a single chat. That's roughly 750,000 words of context. Ten novels. Your company's entire documentation. Finally, no more chunking.
Today in AI:
→ Meta's Manus AI moves from cloud to your desktop
→ Anthropic builds a think tank for AI's workforce impact
→ Prompt of the Day: Deep document analysis at scale
Let's dive in..
🖥️ Manus My Computer: AI Agents Land on Your Desktop
Meta's Manus AI just shipped "My Computer", a desktop app that extends their cloud-based agent to your local machine. It can execute terminal commands, manage local files, and control your apps.
Here's what it can do:
Run terminal commands on your machine (with explicit approval for each one)
Read, analyze, and edit local files without uploading them to the cloud
Launch and control applications directly from the agent
Leverage your local GPU for ML training or LLM inference. That dusty Mac mini becomes a 24/7 assistant
The security model is sensible: every command requires explicit approval. You can hit "Allow Once" to review each operation, or "Always Allow" for trusted recurring tasks.
In one demo, Manus built a Mac app with real-time translation features in about twenty minutes, entirely through CLI commands.
The timing isn't accidental. Following OpenClaw's viral success (196K GitHub stars and counting), local-first AI agents are having a moment. Anthropic acquired Vercept in February. Perplexity announced "Personal Computer" in March. Now Meta's $2B Manus acquisition is joining the race.
For your team: If you're evaluating AI automation tools, the "cloud vs. local" question just got more interesting. Local agents can access sensitive files without uploading them anywhere.
Together with Mintlify
Your Docs Deserve Better Than You
Hate writing docs? Same.
Mintlify built something clever: swap "github.com" with "mintlify.com" in any public repo URL and get a fully structured, branded documentation site.
Under the hood, AI agents study your codebase before writing a single word. They scrape your README, pull brand colors, analyze your API surface, and build a structural plan first. The result? Docs that actually make sense, not the rambling, contradictory mess most AI generators spit out.
Parallel subagents then write each section simultaneously, slashing generation time nearly in half. A final validation sweep catches broken links and loose ends before you ever see it.
What used to take weeks of painful blank-page staring is now a few minutes of editing something that already exists.
Try it on any open-source project you love. You might be surprised how close to ready it already is.
🦞 Tencent's QClaw: AI Agents Go Mainstream in China
Tencent just launched QClaw, a one-click installer that puts OpenClaw AI agents inside WeChat and QQ. Download the app, click a button, and you're running an autonomous agent on your desktop.
Here's what makes this interesting:
It lives in WeChat and QQ. You can send commands to your computer through chat apps you already use
Local-first data. Nothing leaves your device by default
5,000+ skills via ClawHub and GitHub, from file organization to app automation
Multiple AI models built in: Kimi, MiniMax, GLM, DeepSeek (or BYO)
The bigger picture: this is China's AI agent moment. Baidu already integrated OpenClaw into search for 700 million users. Alibaba shipped Qwen3.5 with agent capabilities. ByteDance and Zhipu upgraded their models to support agent workflows. All in the past two weeks.
OpenClaw has 196K GitHub stars. People in China are buying Mac Minis just to run it. Now Tencent's making it accessible to anyone with a WeChat account. That's a distribution advantage nobody in the West has figured out yet.
For your team: If you're watching the AI agent space, China's moving faster on consumer distribution. The tech is similar, but the go-to-market is completely different.
🎯 Prompt Of The Day
When you have 750,000 words of context to work with, the prompt matters more than ever. Here's one for deep document analysis:
I've uploaded [describe your documents — codebase, contracts, research papers, etc. Before answering any questions, read through everything completely. Then provide:
(1) A structural overview — how is this organized and what are the major sections/components?
(2) Key themes or patterns you notice across the full set.
(3) Any inconsistencies, gaps, or areas that need attention.
After this initial analysis, I'll ask follow-up questions about specific sections.The key is asking Claude to read everything before responding. With smaller context windows, models often start answering before they've seen the full picture. With 1M tokens, you can actually use the whole thing.
🐝 AI Buzz Bits
🎬 Hollywood kills ByteDance's global video AI launch. ByteDance paused Seedance 2.0's rollout after viral clips (including Tom Cruise fighting Brad Pitt) triggered cease-and-desist letters from Disney, Paramount, and Netflix. Disney's lawyers called it a "virtual smash-and-grab" of their IP.
⚠️ Musk admits xAI "was not built right" as co-founders keep leaving. The list of departed co-founders keeps growing. SpaceX just poached two programmers from Cursor. The $1.25T merger IPO is still planned for mid-June, but Musk says xAI is "being rebuilt from the foundations." Not exactly confidence-inspiring timing.
🛠 Tool Spotlight
Viktor — AI coworker that lives in Slack and proactively handles marketing audits, lead research, and daily reports.
Logical — YC-backed desktop AI copilot that proactively drafts emails, takes meeting notes, and suggests actions without waiting for prompts.
Query Memory — One API to turn documents and websites into queryable knowledge for AI agents. Handles parsing, chunking, and embeddings so you don't have to build RAG pipelines from scratch.
For a full list of 1500+ AI tools, visit our Supertool Directory
👉 Know someone drowning in AI news? Forward this to them or send your unique referral link
Cheers, Tim
