Models & Benchmarksexplainer
Why AI models hallucinate, and how to contain it
A plain-English explanation of why language models state false things confidently, what research says about the cause, and practical ways to limit the damage.
· 3 min read
03Blog
Writing & Insights
Thoughts on AI, engineering, and building
Practical writing on AI agents, automation, and the engineering behind them, for the people who run the business.
Models & Benchmarksnews-analysis
Reasoning models and test-time compute: what changed, and what it costs
Moonshot's open-weights Kimi K2 Thinking, released this week, shows where reasoning models stand: stronger on research tasks, weaker on software engineering.
· 3 min read
Hardware & Infrastructurebrief
HBM memory is the next AI bottleneck, and SK hynix just showed why
SK hynix's record quarter and its 2026 supply talks, plus OpenAI's memory letters of intent, show high-bandwidth memory as a constraint on AI capacity.
· 2 min read
Models & Benchmarksbrief
MiniMax-M2 sets an open-weights high on a leading index, with caveats
MiniMax's M2 is a sparse open-weights model aimed at coding and agents. What the independent analysis shows and what to check before using it.
· 2 min read
Harnesses & Toolingexplainer
RAG explained: how AI answers questions from your own documents
Retrieval-augmented generation lets a model answer from your files instead of memory. What it is, where it fails, and what to ask before you build it.
· 4 min read
Enterprise Adoptionnews-analysis
AI coding assistants: what the controlled studies and large surveys actually measured
METR's trial, Google's trial, DORA 2025 and Stack Overflow 2025 reach different answers on AI coding tools. Here is what each measured and why they differ.
· 4 min read
Hardware & Infrastructurenews-analysis
AI's real constraint is the grid: what the gigawatt deals and capacity auctions show
Three chip deals in three weeks are priced in gigawatts. Grid forecasts and a record capacity auction show why power, not silicon, now sets the pace.
· 4 min read
Agentsexplainer
Agent, chatbot or script: choosing the right tool for the job
Three terms get used interchangeably. Here is what separates a script, a chatbot and an AI agent, and how to pick the simplest one that works.
· 4 min read
Harnesses & Toolingnews-analysis
Agent protocols in September 2025: MCP gets a registry, AP2 adds payments
The MCP Registry preview and Google's AP2 payments protocol show agent interoperability maturing in layers. What each does and where the gaps remain.
· 3 min read
Enterprise Adoptionexplainer
Anatomy of a workflow audit: deciding what an AI agent should and should not do
Reports of stalled AI pilots point to a design problem. A workflow audit decides what to automate with an agent, with fixed code, or not at all.
· 4 min read
Enterprise Adoptionbrief
Anthropic's Economic Index: businesses use the API to delegate, not to chat
Anthropic's September 15 report shows enterprise API use skews heavily toward automation, and that capability matters more than price in what firms deploy.
· 2 min read
Hardware & Infrastructurehardware-breakdown
GPU, TPU or custom accelerator: who makes AI chips and how they differ
NVIDIA GPUs, Google TPUs and AWS Trainium compared on published memory and bandwidth figures, and what the differences mean for buyers.
· 3 min read