Good Morning from San Francisco,

While you were sleeping, the future snuck a little closer. Figure's new Helix AI system just showed us a glimpse of what's coming: robots that can actually handle your groceries without turning your kitchen into a scene from a slapstick comedy.

Meanwhile, the tech job market here in the US is sending some serious signals. Software developer listings have taken a nosedive to 65% of their 2020 levels. What's behind it? Partly those AI coding assistants everyone's talking about, but also some impressively efficient small teams doing the work of small armies.

Both stories point to the same truth: AI isn't just changing how we work – it's changing how much work we need humans to do. Whether that's good news or bad news probably depends on whether you're a robot or a developer right now.

Stay curious,

Marcus Schuler

Meet Helix: The AI That Makes Robots Think Like Humans

Remember those sci-fi scenes where robots gracefully handle everyday objects? Sunnyvale-based Figure just made that a reality. Their new AI system Helix doesn't just make robots work - it makes them think and learn like humans.

Think of Helix as giving robots a human-like brain. It combines two systems: one for careful thinking and another for quick reactions. Just like how you can both plan your grocery list and catch a falling cup without thinking.

The proof is in the pudding, quite literally. In recent demonstrations, two Figure robots worked together to put away groceries they'd never seen before. One robot could simply be told "Hand the cookies to the robot on your right," and it would understand and execute the task. No pre-programming needed.

What makes this special isn't just what the robots can do, but how they learn. Traditional robots need thousands of repetitions to learn a single task. Helix? It needed just 500 hours of training to handle a wide range of objects and tasks. That's like learning a semester's worth of material in a weekend.

The system's efficiency is remarkable. It runs on basic computer chips and controls everything from individual finger movements to head position. Imagine teaching someone to play the piano, and they pick it up after just watching for a few minutes. That's the kind of leap we're seeing here.

Why this matters:

• We're seeing the first real bridge between sci-fi dreams and practical reality in home robotics
• The breakthrough in training efficiency could mean household robots arrive years earlier than expected
• This technology might finally solve the challenge of robots adapting to messy, unpredictable home environments

Read on, my dear:

• implicator.ai: Meet Helix: The Robot That Learns Like A Human

AI Photo of the Day

Prompt:
old woman crone with blunderbuss in stultifying environment

Tech Jobs in the US Hit Historic Low: AI and Hiring Boom Aftermath

Software developer job listings on Indeed have plummeted to just 65% of their January 2020 levels, marking a dramatic shift in the tech employment landscape. The decline, which represents a 35% decrease and sits 3.5x below the mid-2022 peak, stands in stark contrast to the overall job market's 10% growth since 2020.

This significant downturn in tech hiring can be attributed to several factors, including the widespread adoption of AI coding tools, with 75% of engineers now using some form of AI assistance in their work. Additionally, the aftermath of aggressive pandemic-era hiring and the emergence of highly efficient small teams, like Linear's 25-person engineering team serving 10,000 companies, suggests a fundamental shift in how tech companies approach staffing.

Why this matters:

• The tech sector's extreme boom-and-bust cycle - going from the hottest job market in history to the steepest decline - may signal a permanent transformation in how software is built and teams are structured
• While AI tools drive productivity gains, the trend of small, efficient teams achieving outsized impact could redefine what constitutes an optimal engineering workforce

Read on, my dear:

• pragmaticengineer: Software engineering job openings hit five-year low?

🔧 Cool AI Tool: Making Tech Work For You ⚡

🤖 Meet Proxy 1.0: The AI That Actually Gets Things Done 🎯

Convergence AI's new Proxy 1.0 combines natural language understanding with web automation to perform real-world tasks like booking flights and making reservations - acting less like a chatbot and more like a personal assistant who knows their way around the internet.

AI & Tech News

China Claims Tech Independence Victory as Huawei Founder Boasts Progress

In a high-profile meeting with President Xi Jinping, Huawei founder Ren Zhengfei declared that China's concerns about domestic chip and operating system capabilities have "significantly eased," suggesting a major milestone in the country's push for technological self-reliance.

Microsoft's AI Makes Protein Research 100,000 Times Faster

Microsoft Research has open-sourced BioEmu-1, a groundbreaking AI model that can generate thousands of protein structures per hour on a single GPU - performing in minutes what would take traditional molecular dynamics simulations years to accomplish, potentially accelerating drug development and our understanding of diseases.

Microsoft's Magma Model Makes AI More Physical

Microsoft has unveiled its new Magma AI model that combines object recognition with movement pattern learning, allowing it to control robotic arms and navigate interfaces - marking a significant step toward bridging the gap between artificial intelligence and physical actions.

Meta Rewards Bosses While Workers Get the Boot

Despite soaring profits and a 47% stock surge, Meta is tripling executive bonuses to 200% of base salary while laying off 5% of its workforce and slashing stock options for remaining employees - highlighting a growing divide between C-suite rewards and worker treatment.

Apple CEO Seeks Tariff Relief in White House Visit

Tim Cook rushed to Washington to lobby against Trump's proposed 10% tariff on Chinese imports, as Apple faces mounting pressure from both US trade policies and Beijing's potential App Store investigation, highlighting the tech giant's vulnerable position between its largest market and manufacturing hub.

The AI Race Heats Up: OpenAI Hits 400M Users While Ex-Founder Sulks

OpenAI's meteoric rise continues with 400 million weekly active users and 2 million paying customers, while former co-founder Elon Musk responds by simultaneously suing and attempting to buy the company for $97.4 billion.

AI Decoded 🔓

Your guide to mastering AI tools - no tech degree required.

How I Gamed LinkedIn's Algorithm (Legally)

LinkedIn. The platform is where everyone’s a thought leader, yet nobody has thoughts worth leading with. But fear not – I've spent two decades crafting content that actually gets read. Here's how you can too. And yes, we'll keep the "I'm humbled to announce" posts to a minimum.

The Secret Framework That Actually Works:

Hook Your Reader (Without The Cringe)

1. First line must grab attention (and not with "Hope you're doing well")
2. Break up text into digestible chunks
3. End with a call-to-action that doesn't sound like begging
4. Use emojis like salt: just a pinch, not the whole shaker

Space Rules (Because Size Actually Matters)

1. The first 3 lines determine if anyone reads further
2. Use only 2 lines before "see more" appears
3. Format: [LINE - GAP - LINE]
4. Maximum 62 characters per line (it's science, trust me)
5. Third line: Maximum 50 characters (algorithm magic)

The Five Types of Posts That Actually Get Engagement:

1. Personal stories (but please, no more "I'm humbled to announce")
2. Industry insights (actual insights, not obvious statements)
3. Quick tips (because no one has time for your manifesto)
4. Question posts (real questions, not "Who else agrees?")
5. Data-driven observations (with real data, not "studies show")

Content That Works (And Won't Make You Hate Yourself)

• Tell stories that feel human
• Share failures (LinkedIn loves a good face-plant story)
• Use specific numbers (people can't resist "5 ways to...")
• Ask genuine questions
• Drop surprising stats (but verify them first)
• Keep formatting clean (no bold, no italics, minimal emojis)

Proven Templates That Convert:

Template 1:
"I deleted LinkedIn last year.
Reinstalled it today.
The amount of humble-bragging hasn't changed.
But the strategies that cut through the noise have.
Here's what actually works in 2025:
[Your insights here]"

Template 2:
"Want to know why 90% of LinkedIn posts fail?
They sound like they were written by a corporate bot.
Here's how to write like a human instead:
[Your tips here]"

Template 3:
"3 seconds.
That's how long you have to grab attention on LinkedIn.
I've spent 20 years learning how to do it.
Here's everything I know:
[Your knowledge here]"

Advanced Tips For The Ambitious:

• Write like you talk (unless you talk in corporate jargon)
• Test different posting times (3pm Tuesday isn't magical)
• Engage with comments within the first hour
• Use line breaks strategically
• End with a question that feels natural

The Psychology Behind Viral Posts:

• People engage with authenticity
• Controversy creates conversation
• Solutions beat complaints
• Stories trump statistics
• Vulnerability attracts engagement

Remember: LinkedIn is just a platform. Your content shouldn't sound like a committee of professional buzzword enthusiasts approved it. Write like a human, think like a strategist, and, for heaven's sake, stop starting posts with "I'm honored to announce."

Now go forth and create content that doesn't make people roll their eyes. And if anyone asks where you learned these tricks, just say you're "humbled to share" your success. 😉

(See what I did there?)