The Deep Feed

01 — Lenny's Newsletter

The AI Paradox: Why Automation Demands More Humanity

The shift from software-as-a-service to agent-as-a-colleague

By Lenny Rachitsky · 12 min read

Editor's note: As AI begins to handle the bulk of routine tasks, the value of human judgment moves from execution to orchestration.

The common fear is that AI will hollow out the workforce, leaving a desert of unemployed specialists. But the reality emerging from companies like Every suggests a different trajectory. Instead of replacing people, AI is raising the floor of what a single person can achieve. Dan Shipper, who runs Every as a live experiment in AI-native work, argues that we are entering an era where the most valuable employees are not those who can execute a specific task, but those who can direct a fleet of digital agents. The 'SaaS apocalypse'—the idea that AI will kill software companies—is a misunderstanding of how value is captured. Software isn't dying; it is becoming more integrated into the very flow of human thought and action.

The Rise of the Forward Deployed Engineer

We are seeing a shift in the hierarchy of technical roles. The traditional developer, focused on writing lines of code, is being superseded by the 'forward deployed engineer'. This is a role that sits at the intersection of deep technical capability and high-level product strategy. They don't just build features; they deploy intelligence into specific business contexts. As models like Claude Code become accessible to non-technical staff, the bottleneck moves from 'can we build this?' to 'how do we integrate this intelligence to solve this specific problem?'. The engineer becomes an architect of workflows rather than a writer of syntax.

The only way to stay employed is to ride the models, not fight them.

This transition changes the economics of software. In the old model, you paid for a seat in a platform. In the new model, users may bring their own AI tokens into applications. This shift could actually improve margins for SaaS companies, as they move away from the heavy lifting of hosting general-purpose intelligence and toward providing the specialized interfaces where that intelligence becomes useful. We are moving toward a world where software is built for a partnership between humans and agents, rather than a solo human experience.

The New Work Hierarchy

The Super-Agent: A Slack-integrated intelligence for every employee
Full-Stack Designers: Creative leads who can also direct technical execution
Product Managers: The essential orchestrators of AI-driven workflows
Forward Deployed Engineers: The bridge between raw models and business value

Ultimately, the 'job apocalypse' is a misnomer. Automation does not eliminate work; it changes the nature of what work is worth doing. When the cost of generating a draft or a line of code drops to near zero, the value of the person who knows *why* that draft or code matters becomes immense. We will read more AI-generated text, and we will find it useful, provided there is a human intent guiding the direction.

Key Takeaway

Value is migrating from the ability to execute tasks to the ability to direct intelligent systems.

02 — The Marginalian

The Biological Imperative of Water

Why swimming remains our most direct contact with the natural world

By Maria Popova · 10 min read

Editor's note: A meditation on why the act of swimming is more than exercise—it is an evolutionary homecoming.

There is a fundamental difference between swimming for achievement and swimming for presence. Competitive swimming, with its chlorinated pools and obsession with lap times, is a pursuit of productivity. It turns a natural movement into a transaction. But when we move into the wild water—the rivers, the oceans, the lakes—the objective changes. We stop trying to beat a clock and start trying to exist within a medium. As Roger Deakin observed in his travels through Britain, swimming in nature is a way to find a specific kind of peace that the structured world cannot provide.

The Semi-Aquatic Ancestry

Our connection to water is not merely aesthetic; it is biological. Evolutionary historians suggest that humans spent millions of years as semi-aquatic waders. We are shaped by the water. Our skin, our subcutaneous fat, even the pattern of our body hair suggests a history of movement through liquid environments. When we swim, we are not performing an alien activity; we are returning to a state that our bodies recognise. This is why the sensation of being in the water feels so different from being on land. On land, we are masters of gravity; in water, we are part of the flow.

When you swim, you feel your body for what it mostly is—water—and it begins to move with the water around it.

This metamorphosis is a crossing of boundaries. To enter the water is to leave the world of ambition and desire behind. In the water, survival and presence become the dominant aims. The noise of the world is muffled, replaced by the rhythmic sound of one's own breathing and the movement of the current. It is a rare space where the self can be both lost and found simultaneously.

The Three States of Swimming

Transactional: The pursuit of speed, metrics, and achievement
Transcendental: The loss of self in the rhythm of nature
Biological: The recognition of our creaturely belonging to the earth

To swim in the wild is to accept a certain level of vulnerability. There are no lanes, no timers, and no guarantees of safety. Yet, it is precisely this lack of control that makes the experience so restorative. It forces a confrontation with the present moment that is impossible in the highly controlled environments of modern life.

Key Takeaway

Swimming in nature is an evolutionary homecoming that replaces the drive for achievement with the necessity of presence.

03 — The Marginalian

The Entropy of Stasis

Ursula K. Le Guin on why resisting change is a form of self-inflicted suffering

By Maria Popova · 8 min read

Editor's note: An exploration of why we mistake stability for safety and how that error leads to stagnation.

We have a deep-seated psychological urge to seek equilibrium. We want our lives, our careers, and our relationships to reach a state of permanent stability. However, this desire for stasis is a biological and physical impossibility. The universe is in a constant state of transmutation. To resist this movement is to resist reality itself. Ursula K. Le Guin, in her reflections on change, points out that we often mistake the complacency of equilibrium for actual contentment. We think we are safe because we are still, but stillness in a living system is actually the precursor to death.

The Fallacy of Safety

The central fallacy in our resistance to change is the belief that change will unbalance us. We view change as a threat to our stability, rather than the very mechanism that maintains life. In physics, the end result of total stasis is entropy—the heat-death of the universe. In human terms, the refusal to adapt or to allow ourselves to be altered by experience is a slow descent into a living death. We try to live safely, but safety is an illusion. There is no such thing as a life without risk, only lives where the risks are managed differently.

The more things go on moving, interrelating, conflicting, changing, the less balance there is—and the more life.

Le Guin suggests that we must learn to detach ourselves from our own defensive identities. We hold onto our current viewpoints because they feel solid, but they are merely temporary configurations of energy. When we cling to a version of ourselves that no longer fits the reality of our circumstances, we create suffering. Suffering, in this view, is simply the friction caused by our resistance to the inevitable flow of time and transformation.

How to Embrace Change

Acknowledge that stasis is not equilibrium; it is entropy
View identity as a process rather than a static object
Recognise that resistance to reality is the primary source of suffering
Accept that living fully requires a constant gamble against the odds

To live fully is to accept the gamble. It is to recognise that while change is unpredictable and often uncomfortable, it is the only state in which life can actually occur. The goal is not to find a way to stop the movement, but to learn how to move with it.

Key Takeaway

Resistance to change is not a way to preserve stability; it is a way to invite entropy.

04 — Simon Willison

The High Cost of AI Slop

Why automated 'issue reports' are breaking the feedback loop of software development

By Simon Willison · 5 min read

Editor's note: A warning on how poorly prompted AI tools are polluting the very data they are meant to process.

There is a new kind of noise entering the world of software development: the AI-generated issue report. Armin Ronacher recently highlighted a frustrating trend where users encounter a problem, instead of describing it, feed it into an LLM, and submit the resulting 'summary' as a bug report. The result is a mess of confident inaccuracies. These reports often contain fake minimal reproductions, incorrect implementation strategies, and long lists of error classes that have nothing to do with the actual problem. It is a form of digital pollution that makes the job of an engineer significantly harder.

The Confidence of the Incorrect

The danger of these reports lies in their tone. LLMs are designed to be helpful and assertive, which means they present hallucinations and guesswork with the same authority as factual observations. When a developer receives an issue report, they expect a record of what happened. Instead, they receive a speculative narrative written by a machine that didn't actually experience the error. This forces the developer to spend time deconstructing the report just to find the original, unadulterated truth of the user's experience.

The most frustrating failure mode right now is that people submit issues that are not in their own voice.

This phenomenon represents a breakdown in the feedback loop. Software improves when users provide clear, accurate data about where the system fails. When that data is passed through a 'clanker'—an unthinking, poorly prompted AI—the signal is lost in the noise. We are trading the precision of human observation for the convenience of automated summarisation, and the cost is a massive increase in technical debt and developer frustration.

The Anatomy of a Good Issue Report

The exact command that was run
The expected outcome
The actual outcome observed
The specific error message or log

As we integrate AI into every layer of our workflows, we must be careful not to let it act as a filter that removes the very truth we are trying to capture. The goal of a tool should be to augment human clarity, not to replace it with an automated approximation.

Key Takeaway

Using AI to reword observations into summaries often destroys the specific data required to solve problems.

05 — Simon Willison

The Silicon Scarcity

How the AI boom is driving up the price of everyday electronics

By Simon Willison · 6 min read

Editor's note: The hidden economic consequence of the AI race: the rising cost of consumer hardware.

The massive expansion of AI data centres is having an unintended consequence on the consumer market: it is making electronics more expensive. This isn't due to a general shortage of parts, but a specific reallocation of manufacturing capacity. The memory industry is dominated by just three large players, all of whom have a fixed amount of wafer capacity. They must decide how to split that capacity between different types of memory: DDR for desktops, LPDDR for phones, and HBM for the GPUs that power AI.

The HBM Takeover

High-Bandwidth Memory (HBM) is the essential fuel for the AI revolution. Until recently, it occupied a tiny fraction of total memory production. Now, it is expected to consume 20% of all wafer capacity by the end of 2026. Crucially, a single gigabyte of HBM requires more than three times the manufacturing capacity of a standard gigabyte of consumer RAM. As manufacturers pivot to satisfy the insatiable demand from AI giants, the supply for consumer-grade devices is being squeezed.

AI is killing the cheap smartphone.

This shift is already being felt in the sub-$100 smartphone market. These devices are critical for digital inclusion in regions like Africa and South Asia. As the cost of memory rises, the entry-level tier of technology becomes harder to sustain. The economic priority has shifted from providing affordable computing to the masses to providing high-performance intelligence to the enterprise.

The Memory Hierarchy Conflict

HBM: High margin, high capacity consumption, driven by AI
DDR: Standard desktop/server use, facing capacity diversion
LPDDR: Mobile and low-energy devices, seeing rising costs

We are witnessing a fundamental repricing of the digital world. The 'intelligence' we are building in the cloud is being subsidised, in part, by the rising cost of the hardware we hold in our hands.

Key Takeaway

The massive capacity requirements of AI memory are cannibalising the production of affordable consumer electronics.

06 — The Marginalian

Beyond the 'Why'

John Steinbeck and the power of non-teleological thinking

By Maria Popova · 14 min read

Editor's note: A study in how to see the world as it is, rather than how we want it to be.

Human beings have a relentless compulsion to find purpose in everything. We look at a storm, a tide pool, or a social movement and immediately ask 'why?'. This is teleological thinking—the tendency to explain things in terms of the purpose they serve. While useful for survival, it is a limited way of seeing the world. It forces us to create narratives that often obscure the actual reality of what is happening. John Steinbeck, in his expedition journal *The Log from the Sea of Cortez*, argues for a different approach: non-teleological thinking.

The Tyranny of Purpose

Teleological thinking is often a mask for our own biases. When we ask 'why', we are often actually asking 'who is to blame?' or 'what is the goal?'. This narrows our field of vision. We become so focused on the intended outcome that we miss the intricate, interconnected patterns of the present moment. Steinbeck suggests that science—the pure observation of reality without the interference of interpretation—is the best tool for breaking this habit. By meeting reality on its own terms, free from the need to assign cause or motive, we can see the world more clearly.

The truest reason for anything’s being so is that it is.

This approach is not just an academic exercise; it is a way of living. When we stop demanding that every event have a reason, we become more open to the 'elemental wobbliness' of life. We move from a state of judging to a state of observing. This allows us to see the connections between disparate things—the way a biological process in a tide pool mirrors a social pattern in a city—without the distorting lens of 'purpose'.

Teleological vs. Non-Teleological Thinking

Teleological: Focuses on cause, blame, and intended purpose
Non-Teleological: Focuses on observation, pattern, and existence
The Result: One creates narratives; the other reveals reality

To embrace the non-teleological is to accept uncertainty. It is to sit with the fact that some things simply *are*. This is a difficult state for a mind trained to solve problems, but it is the only way to achieve a truly deep understanding of the world's complexity.

Key Takeaway

True understanding requires us to observe reality as it is, rather than forcing it into the narrow narratives of purpose and cause.