Explainers
Plain-language guides to the things we actually work with: AI agents, security, semiconductors, API economics, and crypto. Every explainer is grounded in our own tooling, datasets, or verified research, and anything time-sensitive is dated.
All explainers
What is agent alignment and why does it matter?
Intent vs instruction, why autonomy raises the stakes, and how alignment is actually tested.
What is an agent orchestrator?
Planning, spawning, model routing, state, and verification gates: the coordinator layer explained.
SEMICON Taiwan 2026 explained
September 2-4, TaiNEX Taipei: dates, venue, scale, themes, and using the exhibitor list remotely.
The npm supply-chain attack explained
The August 2026 keyv/cacheable worm: how it spread, why defenses missed it, and how to check exposure.
Top tips for reducing API costs
Free-model rotation, task-based routing, judging, caching, and batching: what actually worked for us.
DeepSeek API vs OpenRouter API
Direct vs intermediary: pricing, free tiers, rate limits, and when each route wins.
What does 'not your keys, not your crypto' mean?
Custody, counterparty risk, and the tradeoff the slogan leaves out.
How to verify a web scraping provider before you hire one
Probe reports, blocked lists, deliverable definitions: a checklist from 150 live probes.
AI benchmarking in 2026: why single-run scores don't prove anything
Variance, judge instability, contamination, and the multi-run alternative.
Where to go next
- The blog: experiment writeups and verified-data projects.
- Robotics stocks by country: the verified 1,093-ticker index layer.
- The store: the datasets behind everything above.
- Open source: the agent tooling we run in production.