The Deep Feed

01 — Lenny's Newsletter

The End of the Prompt: Astra and the Death of the Interface

How GPT-6 is moving from text-based instruction to genuine computer agency

By Claire Vo · 12 min read

Editor's note: The era of typing into a box is ending; the era of models that actually use your computer is beginning.

For years, interacting with artificial intelligence has felt like shouting into a well. You provide a prompt, you wait, and you hope the model interprets your intent with enough accuracy to be useful. It is a transactional relationship, mediated by the limitations of language. But with the arrival of GPT-6 Astra, that transaction is changing. We are moving away from models that merely predict the next word and toward models that can actually operate the tools we use. This isn't just a marginal improvement in reasoning; it is a shift in how software functions. When a model can navigate a CRM, write code in one shot, or build a 3D asset in Blender without a human guiding every click, the 'prompt' becomes an obsolete concept. We are entering the age of the agent, where the goal is no longer to describe a result, but to delegate a task.

The Capability Benchmark

The true test of these new models isn't found in their ability to write a poem or summarise a meeting. It is found in the 'one-shot' tasks that previously required hours of iterative debugging. Claire Vo’s experience with Astra reveals a model that can tackle product intelligence features and complex hardware hacks that previous iterations simply could not touch. Most notable is the jump in 3D capability. Where earlier models struggled to grasp spatial logic, Astra can generate assets in Blender, moving the needle from simple image generation to actual digital construction. This signals a change in the barrier to entry for technical creation. The bottleneck is no longer knowing how to manipulate the software, but knowing what you want the software to achieve.

The prompt is becoming a relic. We are moving from instructing machines to delegating to agents.

This shift has massive implications for the SaaS industry. If a model can use a browser to perform QA or manage a CRM, the traditional user interface starts to look like a heavy, unnecessary layer. We have spent decades building buttons and menus to help humans talk to machines. If the machine can now 'see' and 'use' the interface like a human does, the interface itself becomes a legacy constraint. We might see a return to more fluid, perhaps even invisible, UIs where the software adapts to the user's intent in real-time, rather than forcing the user to learn the software's specific logic.

Astra's Primary Breakthroughs

One-shot coding for complex product intelligence features
Direct manipulation of 3D software like Blender
Autonomous browser use for quality assurance and CRM management
Hardware integration through CLI and live streaming displays

However, this transition is not without friction. As models take on more agency, the cost of error shifts. When a human makes a mistake in a CRM, it is a typo; when an autonomous agent makes a mistake, it can be a systemic data corruption. The speed and efficiency gains are enormous, but they require a new kind of oversight. We are moving from being 'doers' to being 'reviewers'. The skill of the future isn't just knowing how to build, but knowing how to audit the things that build for you.

Key Takeaway

Software is shifting from tools we use to agents that act on our behalf.

02 — Aeon

The Neoliberal Trap

How the pursuit of market efficiency broke the democratic engine

By Les Coleman · 15 min read

Editor's note: A look at how the economic reforms of the 1980s created the institutional fragility we see today.

Western democracies are currently operating like failing lifestyle businesses. Their leaders, much like middle managers in a declining firm, seem more interested in protecting their own interests and wealth than in pursuing national goals. The result is a cycle of low growth, mounting debt, and a systemic erosion of public trust. To understand how we arrived at this point, we have to look back to the early 1980s, when the neoliberal reforms of Ronald Reagan and Margaret Thatcher fundamentally altered the relationship between the state and the market. This wasn't just a change in policy; it was a change in the very structure of how societies function.

The Structure-Conduct-Performance Model

In corporate finance, the Structure-Conduct-Performance (SCP) model explains how the rules of a game dictate how players behave and what the eventual outcome will be. When applied to nations, the logic holds. The neoliberal era focused on changing the structure: deregulating industries, cutting taxes, and shrinking the state. The theory was that this would unleash market forces and create widespread prosperity. However, the conduct that followed—increased financialisation and reduced regulatory oversight—led to a specific kind of performance: fragility. Instead of a robust, growing economy, we achieved a system where moral hazard is high and the capacity to manage systemic risk is low.

Neoliberalism did not bring prosperity; it brought fragility to previously stable democracies.

Consider the state of public infrastructure. Post-war Western governments invested heavily in the physical networks—roads, energy, water—that underpin economic productivity. The neoliberal shift prioritised privatisation and cost-cutting, leading to a decay in these essential systems. When bridges fail and power grids struggle, it is not an accident; it is the direct consequence of a structural decision to prioritise short-term fiscal metrics over long-term stability. The 'trickle-down' effect promised by tax cuts and deregulation has largely failed to materialise, replaced instead by surging government debt and stagnant per capita growth.

The Consequences of Deregulation

Increased systemic risk due to reduced regulatory oversight
Decay of public infrastructure and shared capital
Rising debt burdens to fund unfunded tax cuts
Erosion of public trust in democratic institutions

The crisis we face is not merely economic; it is institutional. When the state retreats from its role as a coordinator and protector, the vacuum is filled by private interests that do not share the public good as a primary objective. This creates a feedback loop where policy serves the few, further alienating the many, and ultimately destabilising the very democratic structures that allow markets to exist in the first place. Fixing this requires more than just adjusting tax rates; it requires a fundamental rethinking of the social contract and the structural role of the state in a modern economy.

Key Takeaway

The economic policies of the last forty years have traded long-term stability for short-term market efficiency.

03 — Psyche

The Illusion of Choice

Why a deterministic view of behaviour might be more compassionate than free will

By Tara-Lyn Camilleri · 10 min read

Editor's note: A challenge to our most basic assumption about human agency and blame.

When we witness a car accident or a person struggling with addiction, our instinct is to search for blame. We assume that, in that specific moment, the individual could have chosen otherwise. This belief in free will is a cornerstone of our social and legal systems. It allows us to categorise people as 'responsible' or 'reckless'. But this narrative is often a simplification that ignores the complex web of biological, developmental, and environmental factors that actually drive human behaviour. By clinging to the idea of absolute agency, we often miss the actual causes of the problems we see.

The Deterministic Reality

A deterministic view does not mean that outcomes are fixed or that we lack agency. Rather, it suggests that our 'will' is itself a product of prior causes. Our actions emerge from our current biological state, which is shaped by our genetics, our life experiences, and our immediate environment. When we view behaviour through this lens, the focus shifts from moral judgment to understanding conditions. If a person is struggling, we stop asking 'why did they choose this?' and start asking 'what conditions produced this behaviour?' This shift is not just academic; it has practical implications for how we approach everything from criminal justice to public health.

Belief in free will provides a reassuring explanation for misfortune: it allows us to believe that failure is a personal choice rather than a systemic outcome.

The psychological comfort of free will is a double-edged sword. It protects us from the fear of randomness—if everything is a choice, then we can control our own destinies—but it also fuels a culture of intense judgment. Those who endorse free will more strongly are more likely to support punitive measures and less likely to support rehabilitative ones. We treat addiction as a failure of character rather than a complex interaction of neurobiology and environment. This approach is often ineffective because it attempts to solve a systemic problem with an individualistic solution.

Factors Shaping Behaviour

Biological state and genetic predispositions
Developmental history and early life experiences
Immediate environmental triggers and social context
Neuroplasticity and the brain's response to experience

Camilleri suggests a two-week experiment: live as if free will does not exist. Not as a philosophical stance, but as a practical tool for empathy. When you stop defaulting to blame, you begin to see the world with more clarity. You see the patterns of influence and the weight of circumstance. This doesn't absolve people of responsibility, but it redefines it. Responsibility becomes about addressing the causes of behaviour rather than simply punishing the outcomes. It is a move from retribution to resolution.

Key Takeaway

Replacing blame with an understanding of causality leads to more effective and compassionate social solutions.

04 — Psyche

The Presence of Absence

Moving beyond the shallow promises of modern mindfulness

By Nigel Warburton · 6 min read

Editor's note: A critique of the 'be present' mantra and a suggestion for a deeper kind of awareness.

Modern mindfulness culture is obsessed with 'presence'. We are told to focus on the texture of a raisin, the scent of the air, or the sensation of our breath. The goal is to anchor ourselves in the immediate moment to escape the distractions of the digital age. While this can be a useful tool for grounding, it is often a shallow way of experiencing the world. Humans did not evolve to be perpetually stuck in the 'now'. Our survival has depended on our ability to project into the future, remember the past, and ignore the irrelevant noise of the present. To demand constant presence is to demand a state that is fundamentally at odds with how our minds work.

The Haunting of Absence

There is another side to experience that mindfulness often ignores: the reality of what is *not* there. Jean-Paul Sartre noted that when you arrive at a café to meet a friend who does not show up, their absence is not just a lack of a person; it is a tangible event. The absence 'haunts' the space. It colours the light, the sound, and your own internal state. This 'presence of absence' is a profound part of the human condition. To ignore it in favour of a forced focus on the 'present' is to ignore a significant dimension of our emotional lives.

The expectation of a person can make their absence happen as a real, felt event.

Instead of the standard mindfulness exercises, Warburton suggests a different kind of meditation: meditating on absence. Sit with the feeling of someone missing. Acknowledge the void they leave behind. This isn't about wallowing in sadness, but about recognising that our relationships and our losses are just as real as the objects we can touch. The people we love, and the people we have lost, shape our experience of the world through their presence and their absence alike.

Two Ways of Experiencing Reality

Presence: The sensory focus on immediate, tangible stimuli
Absence: The psychological focus on what is missing or expected
The Synthesis: Recognising both as essential to a full life

A truly robust way of being involves the ability to move between these states. We need the ability to focus on the task at hand, but we also need the capacity to carry the weight of our history and our relationships with us. To live only in the 'now' is to live a thin, disconnected life. To live deeply is to acknowledge the full spectrum of what is here and what is not.

Key Takeaway

True awareness includes acknowledging the weight of what is missing, not just what is present.

05 — Not Boring

The World Model Revolution

From generating pixels to simulating reality

By Packy McCormick · 14 min read

Editor's note: A look at the new frontier of AI: models that understand the physics of the world.

For the past few years, the AI conversation has been dominated by Large Language Models (LLMs)—systems that excel at predicting text. But a new category of foundation models is emerging that promises something far more transformative: World Models. Unlike LLMs, which learn the statistical relationships between words, World Models learn from watching the world and the actions taken within it. They are learning the underlying rules of reality—physics, geometry, and causality. This isn't just about making better chatbots; it's about building machines that can understand and simulate the physical environment.

The Death of the Code-Based Interface

One of the most striking developments in this space is Runway’s Solaris. Traditionally, when you want to build an application, you write code that a browser then renders into an interface. Solaris flips this. It is an 'Interface World Model' that generates the application itself, frame by frame, in real time. When you interact with it, the model doesn't execute a line of code; it simply generates the next logical frame of the visual experience. In this paradigm, 'the image is the application'. The distinction between seeing a simulation and interacting with a piece of software is beginning to dissolve.

We are moving from models that write code to models that generate reality, frame by frame.

While Runway is pushing the interface, World Labs is pushing the spatial dimension. Their Atlas model is an 'omni world model' trained on a massive array of data, including text, video, and 3D camera poses. It can reconstruct real scenes into 3D from a single photo and allow for 'pixel-perfect camera control'. This means you can manipulate a digital environment with the precision of a cinematographer, rather than just typing descriptive prompts. It is a leap from 'generating content' to 'constructing worlds'.

Key Players in the World Model Space

Runway (Solaris): Real-time interface and video generation
World Labs (Atlas): 3D reconstruction and spatial intelligence
General Intuition: Research into the fundamental nature of world models

The implications for simulation, robotics, and entertainment are enormous. If a model can simulate a real-time environment with audio and video that responds to user input, we are looking at the foundation for truly immersive digital realities. This is the next stage of acceleration. We aren't just teaching machines to talk; we are teaching them to inhabit the world we live in.

Key Takeaway

The next frontier of AI is not language, but the ability to simulate and interact with the physical laws of our world.

06 — Aeon

The Neuroqueer Lawn

Challenging the Victorian scripts of social propriety

By Siobhan Unwin, Kathryne Ford · 13 min read

Editor's note: A re-reading of history through the lens of emotional excess and social norms.

Imagine a mid-19th-century English garden. It is a scene of perfect, repressed decorum. Then, imagine Hans Christian Andersen, the famed author of fairy tales, mid-meltdown on the lawn of Charles Dickens. This is not a minor social awkwardness; it is a total collapse of the Victorian veneer of respectability. Andersen’s emotional 'overflow'—his intense, messy, and uncontainable sincerity—clashes violently with Dickens's role as the manager of social atmosphere. This scene serves as a perfect entry point into the concept of 'neuroqueering': the practice of challenging the invisible rules that dictate how bodies and emotions are allowed to behave in public.

The Script of Composure

Victorian society, and indeed much of our modern social structure, relies on tightly managed bodies and disciplined emotions. We have unwritten scripts for how to react to bad news, how to express grief, and how to maintain 'decorum'. These scripts are treated as natural and morally superior, but they are actually tools of social control. When someone like Andersen breaks these scripts—when they 'come in too hot' or fail to recover quickly—they are often viewed as a problem to be solved or a person to be pitied. Neuroqueering asks us to stop looking at the individual's failure to comply and start looking at the rigidity of the script itself.

Neuroqueering is not an identity to be claimed, but a verb—a way of seeing how norms are enforced and broken.

The term 'disaster bisexual' is often used online to describe characters who live in a state of perpetual emotional crescendo. While often used affectionately, it points to a deeper tension: the friction between intense, sincere feeling and the social demand for composure. The 'lawn tantrum' is intolerable not because the emotion is bad, but because it exposes the fragility of the social order. It shows that our sense of 'normalcy' is a performance that requires constant, exhausting maintenance.

The Mechanics of Social Enforcement

Expectations of proportional speech and reaction
The demand for rapid emotional recovery
The policing of masculinity and domesticity
The division between private interiority and public display

By applying a neuroqueer lens to history, we can see that the 'problem' was never the person melting down on the lawn. The problem was the expectation that the lawn should remain undisturbed. When we demand that everyone behave within a narrow band of 'acceptable' neurotypical or social norms, we lose the richness of human expression. We trade depth for stability. The challenge is to build a society that can tolerate the overflow without demanding that the person be 'fixed'.

Key Takeaway

Social norms are often just scripts designed to manage emotional intensity, and breaking them is a necessary act of human expression.