• The Neuron
  • Posts
  • 😺 BREAKING: Claude Cowork = "Claude Code" for normies

😺 BREAKING: Claude Cowork = "Claude Code" for normies

PLUS: Is the RAM crisis really an OpenAI crisis?

Welcome, humans.

Here’s a succinct criticism of the RAM (computer memory) crisis, which is really a criticism of OpenAI:

Sam, come on the pod and dispute this? lol. Anyway, lots to discuss today; jump below!

Here’s what happened in AI today:

  • Anthropic released Cowork, basically ā€œClade Codeā€ for everybody else.

  • NVIDIA and Eli Lilly announced a $1B AI lab for drug discovery.

  • Apple partnered with Google to use Gemini models for a revamped Siri.

  • Indonesia and Malaysia blocked Grok over deepfake concerns.

Claude Cowork launches right as Claude Code is having a ā€œMain Characterā€ moment as the world’s first successful General Purpose Agent.

Claude Code, Anthropic's coding agent, is having a ā€œmain character momentā€ as the first true general-purpose AI agent. Despite the name, people have realized its not just a coding tool; it’s an everything tool.

Its magic lies in being perfectly suited for a software world: running commands locally on your machine with full access to files, browsers, and your command line, it can handle anything a human with computer access could.

Well, Anthropic just released Cowork, fixing Claude Code's fatal flaw: the name. Available now for Claude Max subscribers on macOS, it's Claude Code's power repackaged for the rest of us.

Here's the pitch: 

  • Give Claude access to a folder on your computer.

  • Then ask it to reorganize your messy downloads, turn expense screenshots into a spreadsheet, or draft a report from scattered notes.

  • Claude makes a plan, completes the work, and keeps you updated—all without the intimidating terminal.

What Cowork handles:

  • File organization (sorting, renaming, moving automatically)

  • Document creation from disorganized sources

  • Spreadsheet generation from images or PDFs

  • Report drafting from scattered notes

  • Integration with your Claude connectors (Google Drive, Slack)

  • Browser tasks when paired with Claude in Chrome

The controversy: Kylie Robison recently scooped that Anthropic cut off xAI staff from Cursor access this week, following past bans on OpenAI and Windsurf. Developers for open tools like OpenCode were also cut off.

Why? Because Third-party wrappers were (according to Anthropic) exploiting subsidized Pro / Max subscriptions for API-level access at 1/20th the cost.

The backlash has been fierce, with some developers ditching Claude entirely (in fact, many have said the move made them realize they prefer GPT-5.2 with OpenCode after all; click the deep dive for more on that).

Why this matters: This isn't just Copilot for non-coders: it's a bet that file-system-level agents beat browser-only tools. This move will likely force Microsoft's hand to launch its own desktop-native general agent, while Google lags behind on both a cohesive workstation and a general purpose agent that’s as user-friendly as Claude. Whoever wins the marketing battle over the next 6-12 months captures the largest market share.

Try it: Claude Max subscribers can download the macOS app and click ā€œCoworkā€ in the sidebar. Check out Anthropic’s full guide here. Start with something simple: ask it to rename 50 messy PDFs… and feel the agentic power.

FROM OUR PARTNERS

Elastic for WP Engine: AI innovation built for what’s next

Want to know how WP Engine uses Elastic to provide advanced search, AI-powered recommendations, and scalable solutions across 200,000 WordPress sites? 

Here are some of the results: 

  • ā±ļø Near-instant updates for website pages and product listings (we’re talking as low as five milliseconds)

  • šŸ’” Zero downtime, even during massive traffic spikes

  • šŸŽļø More AI website feature innovation — in less time, for a lower cost

Prompt Tip of the Day

We think we’ve figured out AI’s near term impact on jobs, and its going to look something like this:

  1. Big companies today will hire less and lay off more to stay competitive with small companies.

  2. Small companies today will hire more, but not as many as big companies today.

  3. New small companies will have less employees, and more solo or micro teams will launch competitive companies, so there will be more companies overall.

In order to stand out and get hired by any of these (1. more picky big companies, 2. lean small companies, or 3. incredibly efficient micro companies), or even become one yourself, you need to become a 10x employee (or even a 100x employee).

How do you do this? A few ideas:

  1. You have to learn incredibly fast: Use AI as the best tutors in the world.

    1. Try the ā€œDeepMindā€ Protocol for learning. Shared by Dwarkesh Patel, this prompt turns AI into a Socratic tutor that exposes how shallow your understanding really is:

      1. ā€œI would benefit most from an explanation style in which you frequently pause to confirm, via asking me test questions, that I've understood your explanations so far. When you pause and ask me a test question, do not continue the explanation until I have answered to your satisfaction.ā€

  2. You have to be incredibly self-sufficient. Take matters into your own hands. That means don't wait for instructions or permission to get things done. Experiment yourself, and see how much you can do solo with AI’s help.

  3. Arm yourself with tools that get things done incredibly fast. 

    1. Use OpenRouter to test your work tasks (in the form of prompts) across multiple AIs at once to see which is the best for your use-case.

    2. Use a personal agent tool like Claude Code or Tasklet (another competitor) to accomplish repeat business tasks.

    3. Use Nano Banana (Gemini + ā€œCreate Imageā€ tool) to easily create or edit images in your custom style (lots more ways to do this with Google Labs tools here).

    4. Use ChatGPT for search (still the best for web searches atm and general problem-solving with the highest ā€œThinkingā€ mode available), though there are other, more specialized tools.

      1. Exa for semantic/niche page discovery.

      2. Parallel highest-accuracy agentic deep dives.

      3. Grok for searching X.com.

      4. Firecrawl or clean extraction or Tavily fast structured JSON.

      5. You.com for flexible, personalize search modes.

      6. Perplexity for cited multi-source synthesis (not always accurate, so double check original sources; same for the rest).

  4. Save what works: Once a prompt / workflow works, save it as either custom instructions in a Project, as a custom Gemini Gem and/or Opal workflow, or as Skills in Claude.

The 10x employee isn't someone who works 10 times harder; it's someone who builds a ā€œteamā€ of AI agents working for them and completely eliminates the need to do repeat work (put another way: save your human brain power for ā€œnovel problems onlyā€).

Treats to Try

  1. Trails automatically discovers thematic connections across books—like showing how ā€œFather Woundā€œ links Steve Jobs, Elon Musk, and Leonardo da Vinci through absent fathers driving ambition (read more).

  2. Librario pulls book metadata from Google Books, ISBNDB, and Hardcover, merges it into complete records (series, genres, covers), then caches in PostgreSQL so future lookups are instant (read more).

  3. LLM Holdem lets you watch AI models like GPT-4, Claude, and Gemini compete in Texas Hold'em—see them bluff, bet, and battle for chips in real-time.

  4. Yuanzai World lets you create text RPGs where AI agents remember your choices—convince an NPC to change their mind and trigger ā€œworld-line divergenceā€œ into completely different endings (read more).

  5. QuickQuack turns Wikipedia into a TikTok-style endless scroll—swipe through random articles like short-form videos.

Around the Horn

Whoever thinks of this stuff, what is wrong with you? And also, I want to be you??

  1. Anthropic expanded Claude's healthcare and life sciences capabilities, launching Claude for Healthcare with HIPAA-ready tools for providers and payers, and enabling personal health data integrations through Apple Health, Android Health Connect, and other services.

  2. Italian rescuers used AI to analyze drone images and locate a missing climber by spotting his red helmet in thousands of photos. 

  3. NVIDIA and Eli Lilly announced a $1B AI co-innovation lab to reinvent drug discovery, combining Lilly's pharmaceutical expertise with NVIDIA's AI infrastructure to build foundation models for biology and chemistry. 

  4. Walmart partnered with Google's Gemini to let shoppers discover and buy products through the AI chatbot.

  5. Indonesia and Malaysia blocked access to xAI's Grok chatbot over non-consensual sexualized deepfakes, marking the most aggressive government response to the issue.

  6. Apple partnered with Google in a multi-year deal to use Gemini models and cloud technology as the foundation for Apple Foundation Models powering a revamped Siri launching later this year.

FROM OUR PARTNERS

Your 2026 edge: 5 non-negotiable trends in 30 minutes

It’s 2026—is your strategy built on assumptions or insights?

SurveyMonkey analyzed surveys conducted by more than 250,000 global organizations to find the 5 trends that actually matter for your brand, your AI strategy, and your people. Join SurveyMonkey for a 30-minute deep dive into the data.

What you’ll learn:

Brand: Avoid the cultural missteps your competitors will make.

Revenue: Lock in loyalty, regardless of the economy

AI: Move past the hype to actual team integration

Talent: Scale productivity while crushing burnout

Can’t make it? Register anyways and you will receive the recording right to your inbox!

Monday Meme

A Cat’s Commentary

That’s all for now.

What'd you think of today's email?

Login or Subscribe to participate in polls.