Now that coding is gone, what value does our experience still bring? Let me show you by an example.

Turns out there is still lots to do.

Claude (Sonnet) built a corpus of 4,176 review comments and summaries by twelve reviewers, on over 1,700 PRs going back about 20 months. A Fable agent per reviewer pulled out the patterns in what they flagged, with examples: ~300 of them, which another agent compiled into 67 themes. Then I had it turn the themes into checks. Finally it went through every PR merged since the start of September (~500) to work out which checks the version the first reviewer saw would have failed, and whether review caught it. As on many teams, there were already AI review bots on the repo, so “review” in everything that follows means humans and bots together, unless noted otherwise.

That seems worthy of a review bot

As fantastical as Disney World’s imagineered system may seem — and it was, to be clear, predicated on a not-easily-reproducible master plan — the idea of storing utility lines in easy-to-access tunnels is not just for amusement parks. It is possible even in places where most of those lines are currently lying in the dirt, as Prague proved back in the 1970s, when the city decided it was fed up with destroying its precious cobblestone streets every time a buried pipe or cable needed fixing. The city now boasts a sprawling utility tunnel network, occasionally patrolled, evidently, by a couple of robot dogs tasked with mapping out every cable.

Amazing forethought.

](https://apple.news/AlWwvQWbRS8auvbbfAJRY1w?highlight=Somebody%20has%20to%20build%20these%20businesses,%20run%20them%20and%20outcompete%20the%20ones%20already%20there.%20It%20might%20as%20well%20be%20you.)

Somebody has to build these businesses, run them and outcompete the ones already there. It might as well be you.

None of this means nobody ever reviews code. There are absolutely changes where I want another experienced human looking. An example would be a fundamental architectural change. Assuming we did a design session as a wider team, we might want to review the code as a team or agree it was implemented right, or discuss if we want to change anything. Other examples could be something crossing a sensitive security boundary, a change with a huge blast radius, an unfamiliar part of a critical system or simply something where the team says, “I’m not confident about this.”

Review By Exception, not by norm.

To be sure, there will also be a mountain of slop. Of buggy, useless, terrible, offensive software that conforms to no programmer’s sense of aesthetic or purpose and style. But there will also be well engineered, working software that solves real problems, yet hasn’t been reviewed in fine detail by human eyes.

Pretty spot on top to bottom.

Democratic-led commissions have generally found that the industry isn’t doing enough to make broadband universally available, while Republican-led commissions have found the opposite. The Carr FCC maintained that tradition in the latest report.

“Today, we conclude our annual inquiry by finding that deployment of advanced telecommunications capability to all Americans meets the statutory standard of being reasonable and timely,” the FCC report said.

No one would agree with you that their single cable provider is offering speeds that “fast enough”

It’s just a normal Monday morning. You make yourself a nice coffee, open your computer and find yourself with 7 PRs to review. You open the first one: +24506 -3938 lines, accompanied by some AI-generated description of what they’re supposed to do. Somehow, your team has made more changes since Friday than they used to make while you were away for a few weeks.

Sounds about right.

But in 2026, agentic coding practices substantially strengthen the argument for cross-platform desktop apps. Putting aside that the models are stronger at TypeScript than they are at Swift, teams’ feature velocity is increasing. The bottleneck is increasingly verifying the resulting product experiences.

I would argue that having agentic teams strengths the argument for native apps using shared core business logic libraries rather than cross platform run anywhere apps. If the cost of code is nearly zero, the overhead on separate platforms is vastly less. Validation, even for electron, has to be done on each platform. You can’t just run it in Windows and think “yep everything looks great, ship it” So you need platform-specific validation, and writing platform code itself is zero marginal cost. Why wouldn’t you opt in for a better app and experiences.

You don’t need a truck.

We’re doing our best to mitigate the huge increases that are being passed to us, and we’ve been trying to shield our customers from the increases, but the situation has become unsustainable

Remind me the BILLIONS in profits that Apple makes every quarter?

Today, even that loose connection to utility has been broken. I have enormous respect for Qwen, but I very much doubt that a 21GB quantized version of their latest model is more powerful or useful than Anthropic’s latest proprietary release.

If the thing you need is an SVG illustration of a pelican riding a bicycle though, right now Qwen3.6-35B-A3B running on a laptop is a better bet than Opus 4.7!

That is pretty impressive in 18 months.

I was telling my son’s friend about this phenomenon and he responded with the perfect GenAlpha summation:

AI is a D1 glazer, bro.

We need to acknowledge that we’re probably getting the answer we want rather than a cold-hard fact. Not to get too serious but when I read about AI psychosis, I think the overly-confident “You’re a genius” style of reply is the point where it all starts to go wrong.

Hahahaha. A “D1 glazer” indeed.

There’s an uncomfortable parallel between using AI coding tools and playing slot machines28. You send a prompt, wait, and either get something great or something useless. I found myself up late at night wanting to do “just one more prompt,” constantly trying AI just to see what would happen even when I knew it probably wouldn’t work. The sunk cost fallacy kicked in too: I’d keep at it even in tasks it was clearly ill-suited for, telling myself “maybe if I phrase it differently this time.”

For real.

Humans interact with the system almost entirely through prompts: an engineer describes a task, runs the agent, and allows it to open a pull request. To drive a PR to completion, we instruct Codex to review its own changes locally, request additional specific agent reviews both locally and in the cloud, respond to any human or agent given feedback, and iterate in a loop until all agent reviewers are satisfied (effectively this is a Ralph Wiggum Loop⁠(opens in a new window)). Codex uses our standard development tools directly (gh, local scripts, and repository-embedded skills) to gather context without humans copying and pasting into the CLI. Humans may review pull requests, but aren’t required to. Over time, we’ve pushed almost all review effort towards being handled agent-to-agent.

Didn’t know they had coined it “harness engineering”

Eliason hopes their advocacy will have helped transform American cities. “I guess success would be there’s no housing shortage, and we’ve developed super walkable neighborhoods with abundant housing and multigenerational housing,” he said. “We need this broad array of housing types — social housing and market-rate housing and condos and co-ops. Single-stair, at least, helps to unlock some more opportunities.”

Yes!!!!

He pointed out that as average humans, we had never lived the scientific process the way we did with the Covid crisis and the emergence of the vaccine. No wonder everyone from pot-smoking podcasters to scientists had differing opinions. The whole thing was playing out in front of us in real time.

The trials, the debates, the reversals, the uncertainty, the competing claims, the people who were sure they knew what was coming and turned out to be wrong. We watched science happen. It was messy, public, unfinished. And everyone became an armchair epidemiologist overnight.

Living the very messy process right now.

In a news release late Tuesday, Moderna said it was blindsided by the FDA’s refusal, which the FDA cited as being due to the design of the company’s Phase 3 trial for its mRNA flu vaccine, dubbed mRNA-1010. Specifically, the FDA’s rejection was over the comparator vaccine Moderna used.

Fucking ridiculous

We are crossing a threshold in software creation that represents a structural shift in the industry. It’s not just a minor tooling upgrade; it is the end of software as a long-term commitment. We are entering the era of Software on Demand.

For decades, software was something you planned, budgeted, and maintained with heavy “sunk-cost” gravity. Today, with the rise of Agentic CLIs and autonomous coding agents, software is increasingly something you summon. You go from idea to real utility instantly, then decide whether to keep it, evolve it, or simply discard it.

The End of Software as a Permanent Liability

Traditionally, building a tool required weeks of planning and months of engineering. Once built, it became a permanent resident on your balance sheet—demanding maintenance even after its primary usefulness faded.

AI coding tools like Claude Code, Codex CLI, and Cursor are flipping this script. Andrej Karpathy recently described this as a phase shift in software engineering, where developers move from writing syntax to “programming in English.”

This collapses the cost of experimentation. Software is no longer a precious asset; it’s a fluid resource. We can now generate:

• One-off tools for specific data migrations.
• Disposable automations for a single marketing campaign.
• Ephemeral internal apps that only need to exist for a one-week project.
• Rapid prototypes that actually run in production-like environments.

Agentic Tools: From “Assistant” to “Autonomous Engineer”

The industry is moving beyond simple autocomplete. We are seeing the rise of agentic coding tools—systems that don’t just suggest lines of code but understand entire repositories.

Platforms like Replit Agent explicitly market this shift: you describe an idea, and the system builds and deploys an app. It’s effectively an entire team of engineers on demand. These agents can:

1. Understand complex project structures across multiple files.
2. Execute multi-step plans autonomously.
3. Self-correct by running tests and iterating on failures.

“Build the thing. Fix it if it breaks. Improve it if it can."

This is the new directive. No ceremony, no guilt, and—most importantly—no permanent maintenance burden unless you choose it.

The Situational Interface: MCP and Ephemeral Workflows

The Model Context Protocol (MCP) by Anthropic is the “connective tissue” of this new era. By allowing AI models to operate across Slack, Figma, Canva, and analytics platforms, software becomes situational.

Instead of building a rigid, permanent integration between two pieces of software, you can now create a temporary workflow that exists only as long as the intent does. The software adapts to the context rather than living forever as a legacy script.

We are entering the Agentic CLI Era, moving toward command-line agents that execute real engineering workflows end-to-end. Spin up a backend today, generate a data pipeline tomorrow, and kill it next week when the analysis is done.

The New Elite Skill: Curating Software, Not Just Writing It

As tools like the open-source Cline demonstrate, the bottleneck of production is shifting. When an autonomous agent can plan, execute, and browse for solutions, the limiting factor is no longer developer hours.

The bottleneck is now idea quality and judgment.

The craft of the modern developer is shifting from:

• Hand-coding everything $\rightarrow$ Directing intelligent systems.
• Managing Technical Debt $\rightarrow$ Curating results and enforcing correctness.
• Building Infrastructure $\rightarrow$ Deciding what is worth building.

Actionable Takeaway: Start “Vibe Coding”

To stay ahead, stop over-planning for longevity. Start building “micro-tools” for your daily tasks. If a task takes you 30 minutes to do manually, use an agent to build a 2-minute automation. If it stops working next month? Delete it.

In the era of Software on Demand, the most valuable code is the code you aren’t afraid to kill. Software is no longer just infrastructure—it is an on-demand extension of thought.

History doesn’t repeat, but it certainly rhymes in the way it redefines our labor. In the 19th century, the master weaver watched the power loom change the texture of the textile industry forever. Today, the software engineer is watching the LLM do the exact same thing to the fabric of our codebases.

The Shift from Tool to Engine

When the first automated looms arrived, the weaver didn’t stop being an expert on thread, tension, or patterns. However, the physical act of throwing the shuttle—the repetitive, manual labor of the craft—was abstracted away. The weaver became an operator, then a supervisor of a system.

We are currently in the “Power Loom” phase of software development.

For decades, our value was tied to our ability to manually manipulate syntax—the “shuttle” of our trade. We took pride in the speed of our typing and the depth of our knowledge regarding obscure library APIs. AI is moving that manual labor into the background. We are shifting from being the ones who write the lines to being the ones who orchestrate the logic.

The Anxiety of the Artisan

The factory evolution brought a specific kind of fear: the loss of the artisan’s touch. There was a belief that a machine-made rug would never have the “soul” of a hand-knotted one. We hear this now in the discussions about AI-generated code being soulless or prone to hallucination.

The reality? The machine is just another tool in the belt, albeit a much larger one. The factory didn’t kill the need for design; it scaled it. The “coding” part of being a programmer is becoming a commodity, but the programming part—the architecture, the problem-solving, and the deep understanding of the user’s need—is becoming more valuable than ever.

From Syntax to Orchestration

When the factory floor evolved, the most successful workers were those who understood the mechanics of the machine well enough to troubleshoot it, but kept their focus on the output quality.

In this new era, your value isn’t found in your ability to remember the syntax for a map function in five different languages. Your value is in:

• Intent Management: Clearly defining what the system should do.
• System Design: Ensuring automated modules fit into a cohesive, maintainable architecture.
• Verification: Being the final arbiter of truth for what the machine produces.

Actionable Steps for the Modern “Operator”

1. Level Up Your Architecture: Spend less time learning new syntax and more time learning design patterns. The AI can write the code, but you have to tell it where the boundaries are.
2. Master the Prompt, Not Just the Code: Treat your interaction with AI like a senior dev reviewing a junior dev’s work. Be precise, be demanding, and always verify.
3. Focus on the “Why”: Machines are great at the “how,” but they are terrible at the “why.” Double down on understanding the business logic and the human impact of your software.

The factory didn’t end the era of the worker; it began the era of the engineer. We aren’t losing our craft; we’re just trading our shuttles for engines.

Developers can leverage coding agents, including Anthropic’s Claude Agent and OpenAI’s Codex, directly in Xcode to tackle complex tasks autonomously, helping them develop apps faster than ever

🥱

The ultimate result is the ability to create software at scale without directly writing any code, simply by providing strategic direction to a fleet of coding bots. Call it “codeless” software.

This is the next real wave.

mRNA vaccines against high-risk skin cancers appeared to reduce the risk of cancer recurrence and death by nearly 50 percent over five years when compared with standard treatment alone.

Fuck yeah science!

Origin: Kenya
Selection: Peaberry
Notes (card): Dark chocolate, burnt citrus
Vibe: Rich, warming, wintery

This one’s simply called Krampus, which feels right. Every holiday season he shows up uninvited, and somehow this year he brought coffee instead of chaos. This is a peaberry selection from a Kenyan factory, part of the Rutungo farmers group. It’s meant to be rich and warming, with dark chocolate and burnt citrus notes—best enjoyed on cozy winter mornings or at festive gatherings.

Right out of the bag, the aroma is unmistakably chocolate-forward. Fresh, cocoa-heavy, comforting. Something a little different from the usual rotation, and honestly a welcome shift.

Dialing It In

I started this one at -4, which was clearly too fine. The PSI shot straight up to 14 with nothing coming out. When that happens, I usually stop the brew as soon as I see the pressure spike, let it backflush, and then run it again without removing the portafilter. It’s kind of a pseudo pre-soak—lets the puck saturate a bit before pressure ramps back up.

Dialed it back. Still too much. Same issue. By the time the puck was fully saturated, I finally landed around 12 PSI, which felt right. Krampus was harder to dial in than most—true Grind Olympics material.

Iced First

Since it was late afternoon, I decided to save the hot shot for the next day and start with iced.

Over ice, it’s tart but pleasant. I don’t really get burnt citrus in a literal sense, but there’s something leafy and smoky happening—fig-like, with a roasted nut and dark chocolate backbone. There’s a sharpness up front, but it settles into something pretty satisfying.

Hot Shot (Next Day)

The next day, I came back for the hot shot.

Hot, it leans much more almond-roasty and smoky. There’s a definite oolong / hickory vibe—smoke-forward, layered, warming. Interestingly, the citrus that showed up iced doesn’t really come through hot. The espresso version is darker, rounder, and more savory.

Verdict

Krampus is good both ways, but for me:

• Better iced for brightness and contrast
• Hot is smoky, rich, and subtle, but loses some of the edge

Either way, it’s a solid, seasonal curveball—and very on brand for a coffee named Krampus.

Origin: Sierra Mazateca, Oaxaca
Process: Washed
Variety: Typica, Bourbon, Mundo Novo
Tasting Notes: Orange · Fig · Milk Chocolate · Tonka Bean

Sourced from the Sierra Mazateca in Oaxaca, this community microlot highlights coffees produced by women—often uncredited—whose lots scored between 86 and 87. Curated by Osito, Mazateca Mujeres blends small deliveries from female producers, selected for both cup quality and a shared profile. The result is expressive and refined, honoring the women behind some of Mexico’s most vibrant coffees.

Brew Notes:
Right out of the bag, the beans give off a bright, potpourri-like aroma—floral and fruity. In the cup, there’s a vivid orange zest character paired with a rounded milk-chocolate sweetness. The cup stays bright and tangy without tipping into acidity, finishing smooth and clean. Think chocolate-orange candy, but muted and elegant rather than loud.

Verdict:
Bright, expressive, and surprisingly balanced. A standout Mexican coffee that shines hot and showcases a confident citrus-forward profile.