https://icorer.com/icorer_blog/en/categories/ai/Recent content in Ai on MarginaliaHugoen-usWed, 11 Nov 2026 14:12:52 +0800https://icorer.com/icorer_blog/en/posts/ai/sag2-embedding-free-retrieval/Wed, 11 Nov 2026 14:12:52 +0800https://icorer.com/icorer_blog/en/posts/ai/sag2-embedding-free-retrieval/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. The semantic call can wait until after the question arrives. Keep words and counts in the index, and MuSiQue R@10 moves from 54.41 to 86.62 on a 2.36-second build with no model calls. The paper is SAG2: Embedding-Free Multi-Hop Retrieval with BM25 and JEV, by Yuchao Wu at Zleap AI, arXiv 2610.https://icorer.com/icorer_blog/en/posts/ai/jev-harness-design/Wed, 11 Nov 2026 07:16:58 +0800https://icorer.com/icorer_blog/en/posts/ai/jev-harness-design/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. A design note from September 2026 gathers six coding-agent complaints under one sentence. Suppose a language model had no KV cache. What would these agents look like then. The answer contains no stronger model. It contains one thing, someone deciding what goes in front of the model on every turn.https://icorer.com/icorer_blog/en/posts/ai/mopd-token-share/Tue, 10 Nov 2026 07:02:51 +0800https://icorer.com/icorer_blog/en/posts/ai/mopd-token-share/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. TRL’s multi-teacher on-policy distillation example ran 60 steps. Instruction following held 20 percent of the prompts and got 1 to 3 percent of the trained tokens, and it was the only domain that barely moved. Open-MOPD calls this token budget misallocation, and fixing it raised the recoverable headroom from 35.https://icorer.com/icorer_blog/en/posts/ai/ontology-evolution/Mon, 09 Nov 2026 11:08:09 +0800https://icorer.com/icorer_blog/en/posts/ai/ontology-evolution/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. Between December 2025 and November 2026, six papers closed in on one conclusion from three directions. The bottleneck in ontology construction moved from who can write an ontology to who can prove the written thing is right. Every number below goes back to a named section or table. Two minutes twenty-three, voiced in English, captions burned in and switchable in the player.https://icorer.com/icorer_blog/en/posts/ai/evals-honest-numbers/Mon, 09 Nov 2026 07:36:00 +0800https://icorer.com/icorer_blog/en/posts/ai/evals-honest-numbers/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. An eval spends most of its budget on reading outputs and labelling cases. Code is the cheap part. This piece splits the work into four parts and gives each part a number you can recompute. The film runs one minute forty-four, voiced in English, with the captions burned in and switchable in the player.https://icorer.com/icorer_blog/en/posts/ai/decisionrag-retriever-free/Sun, 08 Nov 2026 12:06:43 +0800https://icorer.com/icorer_blog/en/posts/ai/decisionrag-retriever-free/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. The bottleneck in retrieval is not the size of the candidate pool. It is what you let the model see, and in which order. DecisionRAG replaces scoring with asking, and returns a 65.4 top-1 hit over 2.68M passages in 1.7 seconds. It also loses three cells, plainly. English voiceover, 3 minutes 12 seconds.https://icorer.com/icorer_blog/en/posts/ai/graph-serialization-15-formats/Sun, 08 Nov 2026 12:01:41 +0800https://icorer.com/icorer_blog/en/posts/ai/graph-serialization-15-formats/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line. How a graph is spelled into a prompt is a measurable system parameter. It is worth 1.3 to 3.9 points on ontology extraction and 28.4 points on structural reasoning, and the three reasoning tiers of a single model family each prefer a different format, with the most accurate tier burning the fewest tokens.https://icorer.com/icorer_blog/en/posts/ai/agent-economy-project-eve/Sat, 07 Nov 2026 11:25:06 +0800https://icorer.com/icorer_blog/en/posts/ai/agent-economy-project-eve/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: this is not a review of “AI playing a game”. It is the first experiment that splits an agent economy into three billable layers — execution, tooling capital, and the cost of trust — and quotes a price for each: four months of model progress buys 3× the return; swapping the author of a scaffold moves a fleet from losing ships daily to zero losses in twelve hours; and a saboteur who is told to sabotage is almost never caught.https://icorer.com/icorer_blog/en/posts/ai/agent-memory-stack/Fri, 06 Nov 2026 22:12:00 +0800https://icorer.com/icorer_blog/en/posts/ai/agent-memory-stack/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: memory is not a transcript you keep, it is cargo moved between four layers — the low ones are true and hard to use, the top one is easy and easy to get wrong. Dreaming is the worker that moves it while nobody is watching. Note on figures: the diagrams here are redrawn from the article’s own data and structure; its fifteen-odd figures and screenshots are at the source linked at the end.https://icorer.com/icorer_blog/en/posts/ai/agent-native-database/Fri, 06 Nov 2026 11:30:00 +0800https://icorer.com/icorer_blog/en/posts/ai/agent-native-database/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: the database’s main user stopped being a person, so it grew a second leg — one that holds state and provisions on demand, one that supplies context and compiles it ahead of the query. What really changed is the definition of retrieval: from a recall algorithm to an orchestration frame plus a compiled supply.https://icorer.com/icorer_blog/en/posts/ai/seven-layers-agent/Fri, 06 Nov 2026 09:22:00 +0800https://icorer.com/icorer_blog/en/posts/ai/seven-layers-agent/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. Abstract. Most production agent failures do not happen in the model layer. Split the distance from model to result into seven layers — model, instructions and context, tools, memory, loop, evals and guardrails, human signature — and the most common reflex (swap the model, rewrite the prompt) acts on the one layer least likely to be broken.https://icorer.com/icorer_blog/en/posts/ai/recall-company-brain/Thu, 05 Nov 2026 22:42:00 +0800https://icorer.com/icorer_blog/en/posts/ai/recall-company-brain/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: it turns “remembering” into a retrieval contract — the source is stored upstream and never edited, the index and sandbox are rebuildable projections, and the server owns recall while precision and citation go back to the frontier model that asked. Source material: the author’s long post, plus the open skill repository (the engine is not open source; the skills are, MIT, 13 of them).https://icorer.com/icorer_blog/en/posts/ai/deepseek-dsec-sandbox-platform/Thu, 05 Nov 2026 15:28:40 +0800https://icorer.com/icorer_blog/en/posts/ai/deepseek-dsec-sandbox-platform/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: the contribution is not another sandbox runtime. It is a measurement of what agent-training sandboxes actually look like — requests arrive in tens of thousands, CPU usage sits at 5% of what was requested, state lives for minutes to hours, and images are barely reused. Once all four hold, the platform cannot be a container runtime; it has to be an elastic execution platform.https://icorer.com/icorer_blog/en/posts/ai/open-responses/Thu, 05 Nov 2026 08:22:00 +0800https://icorer.com/icorer_blog/en/posts/ai/open-responses/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: it swaps “messages” for bidirectional items, swaps token streaming for state-machine events, and gives extensions a namespaced side channel — so the endpoint and model name become configuration, and the cost moves from adapters to the semantics of individual fields. Read from three places: the specification itself (both the 2026-01-15 and 2026-04-24 versions), its TypeSpec schema and repository, and the docs of two live implementations at Databricks and Vercel.https://icorer.com/icorer_blog/en/posts/ai/ankole-agent-os/Wed, 04 Nov 2026 21:32:00 +0800https://icorer.com/icorer_blog/en/posts/ai/ankole-agent-os/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: a chat box on a single-user harness moves the interface and nothing else. The difference is five things — who is speaking, what it remembers, what survives the process, which machine it acts on, and whether the reply arrived. First, the material boundary. The article’s body is behind an X login; what I could read is its own claim plus the opening paragraph.https://icorer.com/icorer_blog/en/posts/ai/auto-ontology/Wed, 04 Nov 2026 20:41:07 +0800https://icorer.com/icorer_blog/en/posts/ai/auto-ontology/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: Auto-Ontology automates the most expensive month of ontology engineering — it reads your metadata and query history, compiles a business ontology, and hands it to agents through nine read-only tools. What it hands over is a draft: in this codebase, certification is a badge rendered for display, not a gate on the query path.https://icorer.com/icorer_blog/en/posts/ai/one-person-creative-team/Wed, 04 Nov 2026 07:20:00 +0800https://icorer.com/icorer_blog/en/posts/ai/one-person-creative-team/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: the part worth carrying away is the job split. It cuts the creative director role into brand strategy, content strategy, and writing plus ops. The price tags, the product plugs and the growth numbers belong to the sales side of the letter. Dan Koe published a long letter in November arguing that the most lucrative skill has always been capturing attention.https://icorer.com/icorer_blog/en/posts/ai/cognitive-surrender/Wed, 04 Nov 2026 06:54:07 +0800https://icorer.com/icorer_blog/en/posts/ai/cognitive-surrender/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: handing over the work is fine. Handing over the judgement is a different act, and the slope between the two is quiet. English narration, 2 min 41 s. The Chinese cut runs 2 min 33 s with the same scenes and the same numbers. Addy Osmani published a long piece on X in November about keeping critical thinking while working with agents.https://icorer.com/icorer_blog/en/posts/ai/jevact-decision-gate/Tue, 03 Nov 2026 12:27:14 +0800https://icorer.com/icorer_blog/en/posts/ai/jevact-decision-gate/中文版是这篇文章的默认版本。 阅读中文版 This English page stays available at its own URL. The Chinese version is the one maintained first. English readers land here directly. To pin English after switching languages, use ?lang=en on this URL, e.g. this link. An agent’s action should pass through a gate that can check, compute, and send back before anything executes. JevAct splits that gate in two. code settles what tool records decide, a decision model answers only “what did the conversation actually say”, and a boolean rule combines them.https://icorer.com/icorer_blog/en/posts/ai/super-agent-design/Mon, 02 Nov 2026 06:30:00 +0800https://icorer.com/icorer_blog/en/posts/ai/super-agent-design/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: the interesting claim in DeepSeek Harness’s architecture note is not its five modules — modules change with deployment. It is one sentence: sessions should be an implementation detail of the system, not a mental burden on its user. Everything below is that sentence, taken seriously. Published 2026-10-31, the note opens by telling you what it is not: an architecture discussion, not a description of shipping features.https://icorer.com/icorer_blog/en/posts/ai/trusted-mcp-runtime-architecture/Sun, 01 Nov 2026 12:00:00 +0800https://icorer.com/icorer_blog/en/posts/ai/trusted-mcp-runtime-architecture/中文版是这篇文章的默认版本。 阅读中文版 → This English page stays available at this URL; the Chinese version is the one we maintain first. In one line: treat a tool as a content-addressed artifact, run it inside a sandbox, and keep an identity behind it — only then does “trusted” mean anything. That is the architecture; the second half of this article is the implementation. 1. The problem: an agent’s tools are black boxes# An agent’s capability is not bounded by its model.