In 1969, it took a mass of MIT engineers to write the 145,000 lines of assembly code that landed Apollo 11 on the Moon. In 2026, a single operator with a terminal agent can spin up a production-grade full-stack application before their morning coffee goes cold.

We are living through the most violent compression of creative leverage in the history of computing. And yet, the discourse around AI and coding remains frustratingly binary: either "AI will replace all programmers" or "AI is just a fancy autocomplete." Both camps are wrong. The future of coding is not about replacement or augmentation—it is about a fundamental phase transition in what "coding" even means.

In 1937, the economist Ronald Coase published "The Nature of the Firm," asking a deceptively simple question: if free markets are so perfectly efficient, why do companies exist at all? Why don't independent contractors just negotiate with each other on the open market for every step of production?

Coase’s answer was transaction costs. Discovering prices, negotiating contracts, and coordinating labor is exhausting. Firms exist because it is cheaper to internalize that friction through hierarchy than to rely on the open market.

For the last three decades, the enterprise software company has been a monument to this friction. Translating a customer’s raw business pain into deployed infrastructure is an agonizingly slow endeavor. Because the medium of code was historically so rigid, an entire professional class spawned purely to manage this translation layer.

I call them the Delegators.

These are the strategists, the coordinators, the project facilitators, and the alignment-drivers. They do not write the code, and they rarely carry a bag. Their entire economic value—their "moat"—is acting as a human API between the people who have the problems and the people who build the solutions.

But what happens to the human API when the transaction cost of building software drops to zero?

The "SaaSpocalypse" refers to a massive market sell-off in software-as-a-service (SaaS) stocks that accelerated in early 2026. The iShares Expanded Tech-Software Sector ETF (IGV) fell roughly 30% from its September 2025 peak, wiping out over $1 trillion in market value.

But why now?

A deep dive into today's release of Gemini 3.1 Flash-Lite, Google's new model optimized for low-latency, high-volume, and cost-sensitive LLM traffic.

Over the winter break, I finally achieved a goal I’ve been chasing for 15 years. I produced a full-length music album that sounds exactly the way I imagined it.

It’s titled American Intelligence, and it was created with help from Suno AI.

How to authenticate the Google Gen AI SDK.

Large language models (LLMs) like Gemini are incredibly powerful, but they have a fundamental limitation: their knowledge is frozen at the time they were trained. They don't have access to live, real-time information from the internet. This means if you ask about today's news, stock prices, or the weather, they can't give you a current answer.

This is where grounding comes in. Grounding connects the model to external, authoritative sources of information, like Google Search. By giving Gemini the ability to search the web, we can unlock its potential to answer questions about the here and now, ensuring its responses are timely, accurate, and verifiable. 🌐

How to turn Google Docs into speech.

How to build an AI interactive voice agent (IVA) with Gemini Multimodal Live & Twilio.

As artificial intelligence (AI) continues to evolve, one area that remains crucial is intent classification. In the midst of the constant excitement about generative models like GPT-4 and their abilities to produce human-like text, one might wonder, do we still need intent classification? The answer is a resounding yes and is rooted in the deterministic nature of intent classification, a trait that generative AI models, no matter how advanced, cannot currently replicate.