We are living in an era of AI Hype. Every company is shipping features. Everyone is moving fast.

One of the core reasons Biwei and I started Abel is that we are tired of "following" and "mediocrity." We have seen too many products that are "right" but merely "Good Enough." They have no character, no depth. They are stuck in a game of random ping-pong between technical hype and commercial pressure.

This is "The Illusion of Busyness."

"We are adding features," "We are working hard," "We are responding to every piece of feedback..."

The result? A "normal," mediocre startup.

If Abel becomes just another one of them—piling up average features—we have no reason to exist. Tactics don't matter if the paradigm is wrong. Only clear thinking and strong execution on the Regime—Causal AI, the World Simulator—can define us.

The complexity of our product dictates one truth: Focus and Depth are our only moat.

If our way of working is shallow, fragmented, and noisy, our technology will never see through the noise to find the hidden truth behind the moon.

I want to use this memo to recalibrate the most important cultural foundation of Abel. There is only one keyword: Focus.

1. The Violence of Focus

Most people think focus means "locking yourself in a room and working hard."

No. That’s just isolation.

In ancient Chinese, "Focus" (专 * 注) means an intense pouring of energy, a singular, all-in bet. Real focus is violent.

Focus is the violent act of saying "No" to a thousand good ideas.

When Steve Jobs returned to Apple, they had dozens of product lines. He didn’t optimize them. He killed them. He drew a simple 2x2 quadrant on a whiteboard (Desktop/Portable x Consumer/Pro) and cut 70% of the products. That was hundreds of millions in sunk costs. That was the blood and sweat of countless engineers.

That is Focus.

At Abel, we face distractions every day. Investors want more use cases. Customers want custom features. We want to stack more capabilities.

• "If we add this button, users might click it." -> Don't do it. "Might" is an illusion. It distracts us.

• "Competitors have this feature, should we make a lite version?" -> Don't do it. Following is just parroting.

• "LLMs can do this, should we make a wrapper feature?" -> Don't do it. The ceiling of a "wrapper" is not our goal.
  
  

Noise kills signal.

If a feature does not directly serve our core "Causal Reasoning," if it doesn’t increase mathematical depth, reasoning confidence, or interaction joy, it is noise. We must have the courage to subtract.

Do not build more. Build it like a sharp weapon.

2. Focus leads to Depth (The Only Moat)

In this cycle, making an LLM wrapper is easy. Buy Cursor, call a few APIs, write a Prompt, and you have a product. That is why the GenAI market is so crowded—the barrier is low and getting lower.

But this "low barrier" is a trap.

For every "million-dollar wrapper" you see on Twitter, there is a denominator of 10,000 zombie projects that died in the graveyard. They handed their fate to OpenAI’s next update. Sam Altman just has to snap his fingers at DevDay, and that denominator goes to zero.

Peter Thiel said, "Competition is for losers."

It sounds arrogant, but it is the physics of business. If what we do can be replicated by a college student with Cursor in a weekend, we are doomed to a bloody red ocean. That is a low-dimensional war.

Abel refuses to play that game.

Abel’s first-principle moat is not the color of our UI or the layers of our prompts. It is Depth—the depth of Data, Algorithms, Content, and Product.

Without focus, there is no depth. We need Deep Work. Causal AI requires extreme deep thinking.

How do we achieve depth?

You need to dive to the bottom. You need to think about the first principles of data structures and causal graphs. You need to think about how a node's logic holds up in quantization.

You cannot do this if you have 5 meetings a day. You cannot do this if you check your phone every 10 minutes.

At Abel, we don’t reward "hard workers." We reward "deep builders."

• Guard your Focus Time: Ensure at least 3+ hours of uninterrupted time daily. Experience the Flow State. Context switching is expensive.

• Writing IS Thinking: I don't believe in brainstorming; I believe in Writing. Write the Demo, the PRD, the code comments, the flowcharts. Do not call a meeting just to "discuss." Meetings are for decisions, not education. If you can’t write it down clearly, you haven’t thought it through. Writing is the mirror of logic.
  
  

3. Focus leads to Taste and Personality

Focus is the seed of Taste. Taste eventually solidifies into Personality.

Many colleagues think: "Taste is for Designers."

Absolutely wrong. Taste is Systematic Engineering.

Taste requires bandwidth. When you try to stuff 20 features into a quarter, you cannot polish any of them. You accept "Works for me." You compromise with "Good Enough."

When you accept "Good Enough," you kill Taste.

Only when you cut 90% of the noise do you have the bandwidth to obsess over whether an API response takes 200ms or 50ms. Only then can you perfect the elegance of a transition.

This extreme attention to quality, deep thinking, and efficient execution IS Taste.

Look at Linear. Why do the best Silicon Valley teams use Linear over Jira? Linear has 10% of Jira's features.

But its founder, Karri Saarinen, understood one truth: Every extra pixel, every extra click, every millisecond of latency is Cognitive Load.

Taste isn't about making it "Pretty." It's about making the logic Simply Elegant.

Look at Cursor. Why did it replace VS Code for so many senior devs in a month? Is it just an LLM wrapper?

NO. It is a reconstruction of the coding reality.

Cursor deconstructed the task of "Coding."

• It deconstructed Context (turning Prompt Engineering into a topology of the codebase).

• It deconstructed Workflow (Composer + Autopilot).

• It deconstructed Interaction (The fluid "Cmd+K" experience).

It is not a Chatbot. It is a new native architecture.

True Native AI applications must deconstruct the domain to its marrow.

If we are building Causal AI, we cannot just skin an LLM. We must deconstruct "Decision Making" and "Authenticity."

• We need to build a Context Index for decisions.

• We need to anticipate the user's Analysis Workflow.

• We need to reconstruct the causal experience from query to payment.

Abel must embody "Intellectual Elegance."

Our users are here for answers that are elegant, profound, and fun—not for confusion. If our product is full of fluffy language, logical holes, or visual noise, users will not trust our AI to explain the world's cause and effect.

Deep Tech looks like magic, but it’s built on sweat, architecture, and obsession.

The Manifesto

Great products always have Personality.

Mediocre products are like plain water—they try to please everyone, fit every scenario, and end up with no soul.

Great products must be Opinionated. Even Polarized.

Fun. Intellectual. Extreme. Not a people-pleaser. Not trying to be "cute." This is our character.

In a world where LLMs hallucinate, Abel must be "Fun causality that reveals the hidden truth."

To wrap up:

Abel is not for "everyone."

We have chosen the hardest, but most correct path. We choose to build the next-generation engine deep underground while everyone else is panning for gold on the surface.

History is never written by the "followers." History is written by the Outliers—those allergic to mediocrity, obsessed with truth, and paranoid about perfection.

Our success won't be an accident. We are not just winning a market; we are redefining how AI helps humanity understand the world—more structurally, more causally, more intelligently.

So:

Protect your Focus like it’s your life.

Dig Deep until you hit the bedrock of Truth.

Polish your Craft until I can see Abel’s personality.

Don't be a "normal" engineer. Don't be a "good enough" PM. Don't be a "just okay" designer.

In this age of Copycats and Wrappers, let's be the stinging, uncompromising Outliers with a deep soul.

Build deep. Build taste. Build personality.

Stephen & Biwei

From Abel