The Deep Feed

01 — Lenny's Newsletter

The Loop Economy

Why modern companies are abandoning linear growth for self-sustaining cycles

By Lenny Rachitsky · 12 min read

Editor's note: Anish Acharya explains why the old model of building a product and then selling it is dead.

The traditional blueprint for a successful company—build a product, find customers, scale—is being replaced by something more circular. Anish Acharya of Andreessen Horowitz suggests that the most successful modern ventures are not linear trajectories but a series of interlocking loops. In this new model, every action taken by a user or a business process feeds back into the system to improve the next iteration. It is a shift from building static tools to building dynamic engines of continuous improvement.

The Death of the Linear Moat

In the past, a company's competitive advantage, or moat, was often something designed upfront: a patent, a brand, or a specific piece of proprietary technology. Today, moats are discovered rather than designed. They emerge from the way a company's loops interact with user behaviour. When a product uses data from a user's interaction to refine its own performance, creating a flywheel effect, that loop becomes the moat. The more people use it, the better it gets, making it harder for a newcomer to catch up, regardless of how much capital they throw at the problem.

Moats are discovered, not designed.

This shift changes the very nature of product development. Instead of aiming for a perfect, finished state, founders must focus on creating the conditions for these loops to exist. The goal is to build a system where growth and product improvement are the same thing. This requires a different kind of thinking, one that prioritises feedback mechanisms and data-driven refinement over the grand, singular vision of a static masterpiece.

Key components of a loop-based company

Data feedback loops that improve the product automatically
Distribution loops where users become the primary channel for new users
Learning loops that turn operational data into strategic insight

For the agency owner or business leader, this means looking less at what you are building and more at how what you build interacts with itself. If your business model relies on constant, manual intervention to keep the wheels turning, you aren't building a loop; you're building a treadmill. The real scale comes when the system begins to work for you, using its own momentum to drive the next stage of growth.

Key Takeaway

Stop building products and start building systems that feed themselves.

02 — Simon Willison

The Ghost in the Machine

How coding agents are turning Blender into a playground

By Simon Willison · 6 min read

Editor's note: A look at the immediate, practical impact of high-level reasoning models on creative software.

There is a specific kind of magic that happens when a large language model stops being a chatbot and starts being an operator. Simon Willison has recently been experimenting with this by pairing modern frontier models with Blender, the open-source 3D creation suite. The result is not just text on a screen, but actual, editable .blend files and rendered cinematic sequences. The model isn't just describing a scene; it is writing the Python code required to construct it within a complex professional environment.

From Prompt to Render

The workflow is deceptively simple. A user provides a high-level instruction—'render a scene of a pelican riding a bicycle'—and the agent interacts with the Blender Python API to execute the command. This isn't just simple image generation like Midjourney; it is structural creation. The agent understands the geometry, the lighting, and the physics required to make the scene exist in a three-dimensional space. You can then follow up with iterative commands like 'add more flair' or 'make it better,' and the agent modifies the underlying code to reflect those changes.

Models can produce .blend files you can edit in Blender itself.

This represents a massive shift in the barrier to entry for 3D design. Traditionally, mastering Blender required months of learning complex interfaces and specific scripting languages. Now, the interface is natural language. The agent acts as a highly skilled technician who understands the technical minutiae, allowing the human to act as a director. The focus shifts from 'how do I click this button?' to 'what do I want to see?'

Capabilities of AI-driven 3D creation

Direct manipulation of 3D geometry via Python API
Automated rendering of image sequences and movies
Iterative design through conversational refinement

However, this isn't free. While the convenience is high, the cost of these complex reasoning tasks can add up. Using high-end models like gpt-6-astra via API can quickly turn a fun afternoon of experimentation into a significant expense. For the professional, the question becomes one of efficiency: is the speed of natural language command worth the premium over manual technical execution?

Key Takeaway

The future of creative software is not a better UI, but a conversational agent that understands the underlying code.

03 — The Marginalian

The Microscope of the Self

Why divination is actually a tool for psychological clarity

By Maria Popova · 10 min read

Editor's note: A re-examination of Tarot, moving away from fortune-telling and toward self-knowledge.

We have a deep, almost desperate urge to know what comes next. We treat the future as a riddle to be solved, a set of probabilities to be calculated. But the future is not a fixed destination; it is a garden of infinite paths. To believe in a preordained destiny is to surrender the very thing that makes us human: our ability to choose. Maria Popova, exploring the work of filmmaker Alejandro Jodorowsky, suggests that our obsession with predicting the future is actually a misunderstanding of why we use tools like Tarot.

Moving Beyond the Telescope

Most people approach divination with a telescope, hoping to see the distant horizon of their lives. Jodorowsky argues this is a mistake. He is repulsed by the idea that we are prisoners of fate. Instead, he proposes that the true value of Tarot is that it acts as a microscope. It doesn't show you what will happen; it shows you who you are right now. It reveals the parts of your psyche that your conscious mind is too busy, or too afraid, to acknowledge.

The primary purpose of tarot is not to give us a telescope aimed at the frontiers of the future but a microscope aimed at the most elusive intricacies of the self.

When we look at the cards, we aren't seeing a prophecy; we are seeing a reflection. The symbols and archetypes act as a mirror for our habits, our manias, and our unacknowledged desires. By confronting these elements, we can distinguish between what is innate to us and what has been imposed by society, family, or culture. It is a process of stripping away the acquired to find the essential.

How to use divination as a psychological tool

Use symbols to bypass conscious rationalisation
Identify patterns of behaviour that repeat in your life
Distinguish between your true essence and social conditioning

Ultimately, this approach turns an act of 'fortune telling' into an act of intentional living. If we accept that the future is a menu of infinite possibilities, then our task is not to predict which dish we will receive, but to ensure we are the kind of person capable of choosing well. Divination, when used correctly, is not about escaping the uncertainty of life, but about gaining the self-knowledge required to navigate it.

Key Takeaway

Stop looking for answers about what will happen, and start looking for answers about who is doing the happening.

04 — The Marginalian

The Art of Unselfish Listening

Erich Fromm's guide to true human connection

By Maria Popova · 8 min read

Editor's note: A masterclass in communication from one of the 20th century's most important psychologists.

Communication is often treated as a transaction—a transfer of data from one person to another. But if an experience is spoken yet unheard, its reality is halved. The great physicist David Bohm noted that for us to live in harmony, we must communicate in a way where no one is merely defending their own ideas. This is where the work of Erich Fromm becomes vital. Fromm argued that listening is not a passive act, but an art form that requires active, unselfish engagement.

Listening as an Act of Love

Fromm's most striking claim is that understanding and loving are inseparable. To truly understand another person is to reach out to them and overcome the fear of losing oneself in their experience. This isn't erotic love, but a profound empathy where you feel the experience of the other as if it were your own. If listening remains a purely cerebral process—a way to collect information to use later—the door to essential understanding remains closed.

To understand another means to love him.

This level of connection requires a specific set of disciplines. It demands complete concentration, free from the greed of wanting to speak or the anxiety of being judged. It requires a 'freely-working imagination'—the ability to mentally inhabit the world the speaker is describing. It is a difficult, demanding practice that most people avoid in favour of the easier, shallower mode of hearing words without absorbing their weight.

Fromm's rules for mastering listening

Complete concentration: Remove all mental distractions
Freedom from greed: Do not listen just to find an opening to speak
Empathic capacity: Feel the experience as your own
Imagination: Use your mind to build the speaker's world

In an age of constant digital noise and performative communication, Fromm's insights are more relevant than ever. We are constantly 'broadcasting,' but we are rarely receiving. Mastering the art of listening is a way to reclaim the reality of our connections. It moves us from being isolated islands of thought to being participants in a shared, creative movement of human experience.

Key Takeaway

Listening is not the absence of speaking; it is the presence of empathy.

05 — The Marginalian

Regret vs. Remorse

The difference between looking back and moving forward

By Maria Popova · 11 min read

Editor's note: Understanding the distinction between self-destructive rumination and constructive repair.

We all know the feeling of a shattered self-image—that moment when we realise we have acted in a way that contradicts everything we believe about ourselves. In the aftermath, we are often trapped between two distinct emotional states: regret and remorse. While they are often used interchangeably, they represent fundamentally different ways of relating to our mistakes. One is a dead end; the other is a doorway.

The Trap of Regret

Regret is rooted in rumination. It is the desperate, hollow wish for an 'undo' button. It is a backward-looking emotion that focuses on the desire for a different past. Because the past is unchangeable, regret often becomes a turbine of self-destruction. It is a private anguish that serves no purpose other than to punish the self. It is a performance of sorrow that keeps us stuck in the moment of the error, unable to evolve.

Regret is a turbine of self-destruction, remorse an engine of renewal.

Remorse, by contrast, is rooted in reflection and action. It is not just feeling bad about what happened; it is an urgent, directional desire to do something about it. Remorse is about repair. It is the drive to make amends, to heal the person we have hurt, and to transform our understanding of our own character. While regret asks 'why did I do that?', remorse asks 'how can I do better next time?'

Distinguishing the two emotions

Regret: Focused on the past, seeks an undo button, leads to rumination
Remorse: Focused on the future, seeks a redo, leads to repair
Shame: Focused on the self (who I am), whereas remorse is focused on the action (what I did)

The danger lies in the fact that excessive remorse can slip into regret if it becomes too self-focused. If we spend all our energy mourning our own failure, we are not actually doing the work of repair. True morality requires 'unselfing'—moving the focus away from our own guilt and toward the needs of those we have harmed. Only then does the pain of a mistake become the catalyst for growth.

Key Takeaway

Don't use your mistakes to punish yourself; use them to rebuild yourself.

06 — Simon Willison

The Greenfield Fallacy

Why rewriting your code from scratch usually fails

By Simon Willison · 7 min read

Editor's note: A warning to technical leaders about the siren call of the 'total rewrite'.

When technical debt becomes overwhelming, the most seductive solution is to 'burn it all down' and start from scratch. It is a clean, heroic vision: a greenfield project that will finally do things the right way. But in practice, this is almost always a trap. The 'total rewrite' is a classic engineering fallacy that often results in more complexity, more debt, and two broken systems instead of one.

The Moving Target Problem

The fundamental flaw in the rewrite strategy is that the old system cannot simply stop moving. It is the engine of the business. While the new team is building the 'perfect' replacement, the old system must still be updated to handle new requirements, bugs, and market shifts. This creates a widening gap between the two systems. The developers on the old system lose incentive to maintain it, while the developers on the new system are chasing a target that is constantly changing.

The new system is often 80% inactive code that is meant to replace the old system, eventually.

As the rewrite drags on, the pressure to 'just ship it' increases. The result is often a fragmented architecture where the new system only handles a small subset of the original functionality. You end up with a 'janky' old system that no one wants to touch and a new system that is incomplete and fragile. If priorities shift before the replacement is finished, you are left with the worst of both worlds: double the maintenance and half the reliability.

Why rewrites fail

The old system remains a moving target
Implicit knowledge and edge cases are lost in the transition
The 'greenfield' team lacks the context of the original complexity
Business pressure leads to premature, incomplete launches

The better path is not destruction, but migration. Instead of a single, massive leap, focus on shoring up the existing system with automated testing and performing targeted, incremental refactors. This approach respects the complexity of the existing logic and allows for continuous delivery of value. It is less heroic, but it is the only way to actually solve the problem of technical debt without drowning the business in the process.

Key Takeaway

Stop trying to replace your past; start trying to evolve it.