JKJangwook Kimineffloow.hashnode.dev·May 12 · 10 min readMemMachine: Ground-Truth Memory for AI AgentsEvery time an agent summarizes a conversation to save memory, it loses information. That trade-off has been accepted as unavoidable — LLMs produce long outputs, context windows are finite, and token costs are real. MemMachine, presented in arXiv pape...00
JKJangwook Kimineffloow.hashnode.dev·May 11 · 5 min readReFlect: Training-Free Error Recovery for Long-Horizon LLM ReasoningLong-horizon reasoning is where production LLM agents tend to quietly break. A model can produce a plausible-looking chain of thought, accept a wrong intermediate answer, and continue building on that error for every step that follows. By the time th...00
JKJangwook Kimineffloow.hashnode.dev·May 11 · 5 min readClaude Managed Agents: Dreaming, Outcomes, and MultiagentAnthropic released three new Claude Managed Agents features on May 7, 2026: dreaming (a research preview that lets agents learn from their own session history), outcomes (a rubric-based grading system that guides agent behavior toward defined success...00
JKJangwook Kimineffloow.hashnode.dev·May 11 · 5 min readPARSE: Faster LLM Inference via Parallel Prefix Speculative DecodingSpeculative decoding became the standard inference speedup technique through 2024 and 2025. The idea: a small draft model generates a sequence of candidate tokens, and a larger target model verifies them in parallel — accepting the longest valid pref...00
JKJangwook Kimineffloow.hashnode.dev·May 11 · 10 min readZAYA1-8B: Zyphra's Efficient MoE Reasoning Model GuideThe scaling-is-everything story has a new challenger. On May 6, 2026, Zyphra released ZAYA1-8B — an open-weight Mixture-of-Experts reasoning model with 8.4 billion total parameters and fewer than 800 million active per token. On AIME 2025, a benchmar...00
