Blog
Notes and analysis on AI development.
From Inbox to DCF, Why Codex Is Moving Beyond Code
OpenAI Codex use cases now span inboxes, data, finance, QA, app automation, and collaboration, a sign that coding agents are becoming work agents.
Agent memory moved into files, and AMP shows a different path from OpenAI
OpenAI Agents SDK memory and the AMP v0.1 draft turn long-term agent memory into files, Git history, MCP resources, and auditable state.
Why Anthropic is opening in Seoul after 3.5x Claude usage
Anthropic named KiYoung Choi to lead Korea and previewed a Seoul office. The real signal is Korea's 3.5x over-indexing in Claude usage.
CLAUDE.md Became an Attack Surface, TrapDoor's Invisible Instructions
TrapDoor combines malicious npm, PyPI, and Crates.io packages with poisoned AI coding instruction files.
Apps Become MCP Servers, Android Opens a New Door for Agents
Google's Android AI update turns apps into on-device tool surfaces that Gemini and future agents can call directly.
Mistral Picks a Narrow Path Into Industrial Agents With Emmi AI
Mistral AI’s acquisition of Emmi AI shows the LLM race moving into physics simulation, CAD/CAE workflows, and industrial R&D agents.
CUDA 13.3 adds a new tuning lever for LLM inference
NVIDIA CUDA 13.3 targets the lower layers of LLM inference cost with CompileIQ compiler tuning and CUDA Python 1.0.
100 Models Are Heading to the Pixel TPU, but On-Device AI Still Has a Narrow Gate
Google Tensor SDK Beta opens a LiteRT path to the Pixel 10 TPU, pairing 100+ models with the harder problem of shipping on-device AI.
Document clicks are disappearing, and agent GTM needs a new tracking layer
Reo.Dev Agent Intent Gateway shows how developer product evaluation is moving from docs clicks to AI agent queries and MCP calls.
Copilot Memory Turns Agent Memory Into a Control Surface
GitHub Copilot Memory now separates deletion, repository controls, CLI state, and memory scope as coding agents become operational infrastructure.
Safety Warnings Cannot Wait for the Cloud, the Conditions for a Running Agent
Google DeepMind Running Guide agent shows that physical-world agents depend as much on latency, on-device safety, and validation as model quality.
Language Models Need Sleep, Fast Weights Target the Long-Context Bottleneck
Language Models Need Sleep proposes sleep-like fast weight consolidation, shifting long-agent memory from prompt compaction toward model execution schedules.