The Atomic Habit of AI Systems

1% Better: The AI Collaboration Revolution

In addition to this newsletter, I recommend some other great ones. All free. Check them out here.

Friends,

If you haven’t already, you’ll start hearing about MCP when people refer to AI, so I wanted to put together a quick analogous explainer to keep you up to date, just as you started hearing about RAG a year or so ago. I’ll keep it non-technical so it makes sense why it is so important. If you’re a techie, you already know this. If you are not, this will make you sound smartish!

Like ants forming a superorganism or neurons creating thought, the next evolution of AI isn't about individual models becoming smarter, as we hear about every day. It's about many models working together seamlessly. The most effective systems in nature don't operate alone - they synchronize and integrate.

Model Context Protocol (MCP) is changing how AI works by establishing a shared language that allows different AI systems to collaborate without human intervention.

Most improvements in life come from removing friction, not adding motivation. The same applies to AI. Today, you juggle between AI tools manually, perhaps drafting in one system, editing in another, and sending elsewhere. With MCP, this friction disappears. The system handles the transitions while maintaining full context.

The power of MCP comes from three core capabilities:

• Context preservation: Each AI inherits the complete history and knowledge needed to continue where another left off.

• Role clarity: Each AI understands its specific responsibility within a larger workflow.

• Handoff efficiency: Work automatically transfers between specialised AI tools, eliminating the need for repetitive explanations.

OpenAI's ChatGPT Operators (Premium) feature demonstrates MCP principles in practice. When applying for a visa, the system orchestrates these steps instead of manually coordinating between search tools, document scanners, and form fillers.

If you want to deep dive - this is a good video to understand further:

Claude functions similarly within multi-agent systems, summarising documents before forwarding them to specialised agents for fact-checking and tone adjustment. Each AI comprehends what it receives, the transformation it must perform, and the next destination for the results.

Some systems even leverage blockchain technology through Thirdweb's MCP Server, creating digital assembly lines where independent AIs collaborate without centralized control.

This progression matters because it follows the same pattern as all meaningful improvements: Small daily gains in productivity compound as systems handle increasingly complex workflows without intervention. For developers, building becomes about connecting specialised tools rather than creating monolithic systems. The means that work shifts from isolated AI interactions to orchestrated collaborations happening invisibly in the background.

Just as in life, we don't rise to the level of our goals - we fall to the level of our systems. MCP creates systems where AI tools rise together. The 1% improvement that changes everything isn't a single smarter AI. It's the invisible protocol that allows many AIs to work together. I think a medical example sums this kind of integration up and what you will seamlessly see in the future. For example, we might see healthcare diagnosis systems where one AI analyses medical images, another reviews patient history and lab results, and a third combines these insights with the latest research to generate comprehensive treatment recommendations, all flowing seamlessly through MCP without requiring doctors to manually coordinate between tools or restate information. This collaborative approach would dramatically reduce diagnostic time while increasing accuracy, as each specialized AI contributes its strengths to a unified patient care workflow. It’s what I wished for on my last hospital stay!

Stay Curious - and don’t forget to be amazing,

Rahim

Here are my recommendations for this week:

One of the best tools to provide excellent reading and articles for your week is Refind. It’s a great tool for keeping ahead with “brain food” relevant to you and providing serendipity for some excellent articles that you may have missed. You can dip in and sign up for weekly, daily or something in between - what’s guaranteed is that the algorithm sends you only the best articles in your chosen area. It’s also free. Highly recommended. Sign up.

Now

1. The 14 most life-changing nonfiction books of all time - A great list from memoirs and history books to genuinely useful self-improvement. These are my additions to the list: Man’s Search for Meaning – Viktor E. Frankl, The Body Keeps the Score – Bessel van der Kolk, Atomic Habits – James Clear, The Power of Now – Eckhart Tolle, Range – David Epstein, Factfulness – Hans Rosling, The Second Sex – Simone de Beauvoir, Deep Work – Cal Newport, The Prophet – Kahlil Gibran, Grit – Angela Duckworth, Quiet – Susan Cain, Meditations – Marcus Aurelius, The Untethered Soul – Michael A. Singer, Outliers – Malcolm Gladwell, The Obstacle Is the Way – Ryan Holiday.

2. Bill Gates: Within 10 years, AI will replace many doctors and teachers—humans won’t be needed ‘for most things’ Bill Gates predicts that AI will replace humans within ten years in most tasks, including teaching and medical advice, making “free intelligence” widely accessible. He envisions AI addressing fundamental needs such as education, healthcare, food production, and logistics while acknowledging that some human roles, like sports or personal connection, persist. Gates challenged OpenAI to develop an AI capable of excelling in the AP Biology exam, expecting it to take years; they completed it in months. He describes this as the most significant technological leap since the GUI. While optimistic, he cautions against misinformation and urges young entrepreneurs to establish AI-first businesses.

3. What is the meaning of life? 15 possible answers – from a palliative care doctor, a Holocaust survivor, a jail inmate and more. I found this incredibly rewarding to consider. Common themes include love, connection, presence, and purposeful struggle. Kathryn Mannix, after decades in palliative care, says meaning crystallises in mortality: people value relationships, not success. Susan Pollack, liberated from Bergen-Belsen at 14, credits a soldier’s kindness as her turning point. Bill McKibben reframes purpose as preserving the planet’s future. Oliver Burkeman focuses on feeling fully alive. Despite vast differences in background, nearly all responses converge on a simple truth, meaning is made, not given, and emerges through how we respond to life, not define it. also: A Surprising Route to the Best Life Possible (must read!)

4. Are you cleaning your water bottle enough? YUK! Reusable water bottles can harbour millions of bacteria within hours. A Purdue University study found bacterial counts soaring from 75,000 to over 2 million per ml in just one day. Most contamination arises not from tap water but from users through saliva, unwashed hands, or sugary drinks. Bottles have even tested positive for antibiotic-resistant strains like Klebsiella grimontii. Proper cleaning is uncommon: 15% of people never clean their bottles, and many merely rinse. Cold water isn't sufficient. Experts recommend hot water (60°C+), detergent, air-drying, and cleaning every part after each use. Glass or stainless steel is safer than plastic due to chemical leaching.

5. 10 Pieces Of Honest Career Advice No One Tells You - Many professionals in their 30s and 40s reach a turning point: reassessing careers that once felt right. 10 hard-earned lessons: titles are temporary, visibility beats hard work, and burnout isn’t a badge. Success means aligning work with purpose, not just chasing promotions or prestige. Interesting insight: your dream job at 25 might not suit who you've become at 35. Fulfilment is designed, not stumbled upon, and clarity, not control, drives real progress. Career growth is nonlinear; it requires courage to evolve, pause, or pivot. Meaningful success comes from matching work to purpose, not chasing external validation.

Next

1. Scarcity and Abundance in 2025 - Alex Danco examines how the abundance of AI is reshaping software, economics, and company structures. Drawing from previous essays, he argues that we have entered a new “Agent S-curve” era where code behaves more like labour than capital. Value no longer resides in codebases but in how agents perform tasks. Incumbents overwhelm users with unnecessary features, while agile startups thrive by offering precise, task-focused tools. The latest Y Combinator batch is reportedly generating more revenue than any previous cohort, driven by AI’s low-cost, high-skill "workforce." Danco observes that the most transformative work occurs between firms rather than within them, especially through on-chain, stateless AI agents.

2. Has the Decline of Knowledge Work Begun? The unemployment rate for graduates has risen faster than for other workers in recent years. How concerned should they be? White-collar unemployment is increasing more rapidly than in other sectors, driven by AI adoption, corporate restructuring, and cuts to federal spending. Since 2022, the unemployment rate for graduates has risen by 30%, outstripping the general increase of 18%. Companies like Starbucks, Wells Fargo, and Johns Hopkins are reducing corporate roles, citing efficiency and funding pressures. AI coding assistants have enhanced productivity by over 25%, particularly for junior developers, indicating a diminished value placed on experience. Analysts warn that this shift may permanently alter the landscape of knowledge work. The era of stable, well-paid white-collar jobs may be yielding to leaner teams, automation, and a redefined valuation of skills.

3. The Bitter Lesson: Rethinking How We Build AI Systems - In AI, raw compute outperforms clever engineering. Richard Sutton’s "Bitter Lesson" is unfolding in real time: scaling computing, rather than writing more rules, leads to stronger systems. Rule-based and prompt-engineered agents struggle with edge cases. In contrast, reinforcement learning (RL) agents that explore multiple paths and engage in self-learning perform better. DeepMind's AI-designed proteins and Claude’s 3.7 agentic behaviour exemplify this. The most significant shift is strategic investment in scalable computing, as opposed to complex code. Engineers must design systems that learn through parallelisation and feedback loops. Compute-led RL agents reveal better solutions than humans could ever script manually.

4. Scientists Just Transplanted a Pig Liver Into a Person for the First Time - Scientists in China successfully transplanted a gene-edited pig liver into a brain-dead human for the first time. The liver began producing bile within two hours and functioned for ten days without rejection or inflammation. The organ was sourced from a Bama miniature pig that had undergone six genetic edits to prevent immune response. Although the patient had a functioning liver, the pig organ performed essential tasks such as bile and albumin production. Currently, over 104,600 individuals are on transplant waitlists. This breakthrough suggests that pig livers could provide temporary life-saving support while patients await human organs or recovery.

5. How Software Engineers Actually Use AI - WIRED surveyed 730 developers about AI in coding. Most see AI not as a job threat, but as a tool for automating tedious tasks, leaving creative and debugging work to humans. Some view AI as a “hyperefficient intern" without context or judgment. Notably, when asked to analyze the survey, ChatGPT fabricated quotes, misinterpreted categories, and omitted job types, highlighting AI’s limitations. While AI reshapes coding, engineers remain essential. The message is clear: embrace AI as a collaborator, not a replacement, or risk being left behind.

Free your newsletters from the inbox: Meco is a distraction-free space for reading newsletters outside the inbox. The app has features designed to supercharge your learnings from your favourite writers. Become a more productive reader and cut out the noise with Meco - try the app today

If you enjoyed this edition of Box of Amazing, please forward this to a friend. If you quote me, or use anything from this newsletter, please link back to it.

If you are interested in advertising in this newsletter, please see this link