Blog
Hermes-Agent vs OpenClaw: The New Frontier in AI Agent Frameworks
Hermes-Agent and OpenClaw both aim for persistent assistants that learn over time—but with learning-first vs ecosystem-first architectures. Compare memory, performance, enterprise use, and developer experience in 2026.
Revolutionizing Finance: AI Applications in Hong Kong's Financial Sector
How AI is transforming Hong Kong’s banking, trading, fraud detection, and RegTech—and what Fintech 2030 means for institutions and consumers.
Revealed: Claude Code's 510,000-Line Leak Exposes the Hidden 60/40 Secret of AI Engineering Success
Claude Code's accidental leak exposed ~1,902 files and 510,000+ lines—revealing a practical 60/40 split between model capability and engineering harness, plus system prompts, dual-model Auto mode, and a selective memory approach.
Evolving AI: Agentic Variation Operators (AVO) — How Machines Are Learning to Improve Themselves
Agentic Variation Operators (AVO) is a new approach where AI agents take part in their own evolution—sampling weaknesses, generating improvements, and evaluating what works.
The Evolution of Legal AI Tools: Transforming the Legal Industry
Explore leading legal AI tools worldwide and compare three Hong Kong-focused solutions: Curiosity by AlphaMatch, Lexis+ AI, and GenieAI.
CLI-Anything: Why a GitHub Repo with 21K Stars is Changing How AI Agents Use Software
HKUDS CLI-Anything wraps real apps like GIMP and Blender in production-grade CLIs with JSON output and tests—so AI agents use structured commands instead of brittle GUI automation.
