Anthropic's $30B run-rate makes the Claude Code design sprint look like table stakes.

From the editor

Today's items form a coherent arc. Anthropic's revenue tripling to $30B in eight months is the headline, but the real story is what that capital buys: Claude Code shipping UI generation and MCP Inspector maturing into a serious tool. Meanwhile, Lovable's CTO is essentially describing the world Anthropic is building toward, and the ARC methodology critique lands as a useful reminder that safety evaluation rigor needs to keep pace with commercial velocity. The macroagent ontology paper is early, but the framing matters more than the timing.

TL;DR

Anthropic's annualized revenue hits $30B, tripling in eight months on compute deals.
Claude Code adds a /design command, closing the gap between concept and shipped UI.
Lovable's CTO says the next SaaS era belongs to MCP-native, agent-ready applications.

7 curated itemsscroll for the brief

Releases

What shipped · 3 items

v2.1.247

New release of Claude Code with bug fixes, feedback tool improvements, API integration updates, auto-mode enhancements, and cost optimization.

Claude Code

Claude Code วาด UI ได้แล้ว, /design เปลี่ยนวิธีทำ frontend

Claude Code introduces UI design capabilities via the /design command, letting developers go from concept to code without leaving their CLI workflow.

dev.to

inspector: 2.4.0

MCP Inspector 2.4.0 ships with bug fixes, new features, authentication improvements, and web UI updates.

MCP

Reading

Long-form signal · 3 items

Anthropic says run-rate revenue passed $30 billion, up from about $9 billion eight months earlier

Anthropic reports its annualized revenue has tripled to over $30 billion in just eight months, driven by massive compute monetization with partners like Google and Broadcom.

dev.to

Where Did D Go? A Gap Between ARC's Motivation and Its Formalism

A critical examination of a gap between ARC's stated motivations and its formal evaluation methodology, touching on adversarial robustness, trojan attacks, and AI safety evaluation metrics.

LessWrong

Lovable CTO: The Future of SaaS Is Apps That Agents Can Use

The CTO of Lovable argues that the next era of SaaS will center on MCP-compatible applications designed for agent access, reshaping how software products are built and monetized.

Latent Space

Discussions

Where it heats up · 1 item

[Macroagents] 1. The Macroagent Ontology

A new conceptual framework for thinking about macroagents as a distinct ontological category in AI systems, exploring their optimization properties and alignment implications.

LessWrong

Always at hand

Reference links you keep open