Draft records passing an evidence gate into retrievable knowledge cards, then returning through a review and retirement loop

AI Knowledge Base Workflow: Turn Notes Into Verified, Reusable Capability

Growing Bigger, Getting Less Useful We’ve reached layer three. Information has been captured and denoised; records have been written down and polished into semi-finished products — now the question is: how do you turn these semi-finished products into actual knowledge? Let’s start with a definition. Knowledge is structured, repeatedly reusable material relevant to you: a mental model, a handful of skills, a methodology, along with your judgment, positioning, and values. Its keyword is reusability, and it solves your own problems. ...

July 11, 2026 · 11 min · 2202 words · Xinwei Xiong
Text, voice, screenshots, code changes, and decisions moving through capture, clarification, and review toward a verified knowledge card

AI Note-Taking Workflow: Turn Fleeting Inputs Into Verifiable Records

The Semi-Finished Product Filed Under “Knowledge” In most people’s mental model, notes only have three tiers: see information → turn it into knowledge → use it to create. The act of “recording” in between gets quietly filed under “knowledge.” But as I said in the overview, records deserve to stand alone as their own layer. Because it’s an independent intermediate form: it’s relevant to you, but not necessarily useful forever; it might just be something you’ll need someday, or something you’re using right now to clarify your own thinking. That kind of thing doesn’t yet qualify as knowledge — only what’s structured for repeated future reuse counts as knowledge. ...

July 11, 2026 · 10 min · 1923 words · Xinwei Xiong
Sources passing through relevance, privacy, and verification gates before AI-assisted processing and entry into the records layer

AI Information Filtering Workflow: Capture Signal Without a Noise Archive

The Default State of Information Is Noise The previous essay laid out the framework: information, records, knowledge, and creation are four distinct stages. This one deals with only the first — information. The single most important thing to understand about information is this: its default state is noise. We have a natural greed for information. See a good article, want to bookmark it. See a great quote, want to save it. See a reading list someone recommended, want to add it to your queue. Every act of “saving” gives us a small illusion of “I’m making progress.” But saving, at its core, is just moving information from someone else’s warehouse into yours — it hasn’t gone through any processing by your own machine. ...

July 11, 2026 · 9 min · 1889 words · Xinwei Xiong
Information crossing intake gates into records, verified knowledge, audience-facing creation, and a measured feedback loop

AI Knowledge Workflow: From Information to Records, Knowledge, and Creation

Four Names for Four Different Kinds of Work My notes used to grow in one direction: inward. Links entered, fragments accumulated, folders changed names, and the archive became heavier. I mistook possession for processing. AI made that mistake cheaper to repeat. A model can generate, summarize, classify, and reformat text quickly, but speed does not turn a source into evidence, an observation into knowledge, or a draft into something I should publish. It can make the warehouse larger without improving the machinery. ...

July 11, 2026 · 10 min · 1918 words · Xinwei Xiong
A verifiable network connecting an author, original work, independent evidence, and reader communities

Off-Site Trust for GEO: Identity, Evidence, and Ethical Distribution

The answer first: trust is evidence, not a platform trick A technically healthy page can be crawlable, clear, and quotable and still receive no visible citation from an answer engine. That does not prove a hidden “trust gate” rejected it. The page may never have been selected, another source may have matched the prompt better, the answer may have absorbed its facts without linking it, or the platform may have behaved differently on that run. ...

July 11, 2026 · 12 min · 2453 words · Xinwei Xiong
open-lovable architecture from web scraping through code generation to dual-sandbox preview

Inside open-lovable: Search, Text Protocols, and Sandboxes

open-lovable begins with a compact promise: give it a website, get back a React application you can keep editing. The generated page is the visible trick. The more useful subject is the plumbing underneath—how scraping, model output, and untrusted code execution are joined without surrendering the whole process to the model. First, pin the clock. This article audits firecrawl/open-lovable commit 69bd93bae7a9c97ef989eb70aabe6797fb3dac89 , dated November 19, 2025. “In the project” below means that exact revision. “The platform today” refers to current vendor documentation. A fixed commit is a small discipline, but it prevents later product features from being smuggled into an older implementation. ...

June 29, 2026 · 9 min · 1809 words · Xinwei Xiong
The Super-Individual Stack: AI-Native Product Directions and Solo Builder Ops in 2026

The Super-Individual Stack: AI-Native Product Directions and Solo Builder Ops in 2026

“Software is eating the world.” — Marc Andreessen, 2011 “Now AI is eating software—and the question for the rest of us is: what’s left for one human, alone, in front of a screen?” — me, asking myself one night in 2026. Prologue: How Big Does One Person Need to Be? In February 2026, I ran my first complete overnight agent. I set a prompt, dropped it into Claude Code in a loop, and went to sleep. At 7 a.m. the next morning, what I saw on the screen was: 6 commits, 4 PRs, 3 auto-rolled-back failures, and a research brief I hadn’t even read myself. ...

June 24, 2026 · 21 min · 4314 words · Xinwei Xiong
Editorial architecture diagram of Relay with a coordinator, five domain agents, safeguards, browser delivery, and audit boundaries

Relay Agent Architecture in 2026: A Local Implementation Audit

An architecture diagram is a promise. A code audit asks which parts of that promise have acquired weight. My first version of this article treated Relay as a public open-source proposal whose Agent layer had not yet been built. Both claims are now wrong. As of July 31, 2026, the public GitHub URL previously cited by this article returns 404, while I can inspect a private local checkout. This piece is therefore a private/local implementation audit, fixed to local commit 22586e17ccd43cfaff0512511e71a100c5341608. Readers should not assume that repository or commit is publicly downloadable. ...

June 24, 2026 · 11 min · 2298 words · Xinwei Xiong
A wide schematic of context engineering: the Write / Select / Compress / Isolate pillars feeding an AI, a laptop with notes, and a local-first world line

Context Engineering: The New Foundation for AI Agents

Context engineering is the set of strategies for curating, ordering, and evicting the tokens that enter an LLM’s context window on each inference call—system instructions, retrieved documents, conversation history, tool definitions, and memory. The short version is this: prompt engineering optimizes the wording of an instruction; context engineering optimizes the wiring of the whole window. Anthropic has articulated this distinction in its engineering guidance, and Karpathy helped popularize the term. The rest of this article takes the emerging discipline apart. ...

June 22, 2026 · 18 min · 3655 words · Xinwei Xiong
A technical diagram with a tiny agent loop at the center, surrounded by concentric rings of the eight pillars: orchestration, context, memory, tools, reliability, evaluation, cost, governance

Agent Engineering Harness: The Eight Pillars Behind the 98.4%

A small model loop becomes a system only after we decide what it may remember, touch, spend, and survive. An agent demo can be a loop around an API call. A production agent is the surrounding discipline: state, tools, permissions, recovery, evaluation, and cost. This article offers a map of that discipline rather than a framework tutorial. Each pillar is described through the gap it fills, a minimal implementation, and the boundary where it stops helping. ...

June 17, 2026 · 20 min · 4223 words · Xinwei Xiong
Files, memory, runtime boundaries, and evaluation for continuous agent identity

Agent Identity: From Locke to OpenClaw

A philosophical boundary and an engineering practice for continuity in AI agents Start by Making “Identity” Small Enough to Engineer Agent amnesia first damages the cost of cooperation, not the illusion of personhood. A strong session does not guarantee that the next one will preserve the same judgment. Users repeat preferences, teams restate constraints, and the system struggles to explain where an earlier decision came from. Long-term trust rests on these unglamorous forms of continuity: what the agent remembers, what it forgets, why it changed, and who approved the change. ...

April 5, 2026 · 16 min · 3207 words · Xinwei Xiong
How to Maintain the Weight of Self in an Age When You Are No Longer Needed

Maintaining Self-Worth in the Age of AI

Pascal wrote in the 17th century: “All of humanity’s problems stem from one thing: man’s inability to sit quietly in a room alone.” Three hundred years later, I thought of this sentence late at night in Lhasa, and added one more: They’re not sure who exactly is the one sitting in that room. Introduction: The 3 AM Emptiness For a while, I woke up almost every day at 3 AM. ...

April 4, 2026 · 10 min · 1963 words · Xinwei Xiong
A quiet archive of linked memory cards illustrating Mem0 OSS v3 retrieval

Mem0 OSS v3 in Practice: Memory Architecture, Retrieval, and Trade-offs

This project note is part of my attempt to understand open-source AI systems by building with them, reading their migrations, and writing down where the abstraction holds—and where it leaks. Project learning list The problem is not remembering more An LLM can carry a conversation while the relevant messages still fit inside its context window. That is short-term continuity, not memory in the human sense and not durable application state. When the session ends, the model does not naturally retain that a user prefers terse answers, changed jobs last month, or abandoned an earlier plan. ...

May 9, 2025 · 12 min · 2421 words · Xinwei Xiong
Microsoft UFO desktop agent architecture

Microsoft UFO³ Explained: From Windows Desktop AgentOS to a Multi-Device Agent Galaxy

Why UFO matters Desktop automation has always had an awkward boundary. Scripts are fast and precise when an application exposes the right interface. GUI automation is broad, but a changed label, a delayed window, or a custom-drawn control can break an otherwise sensible workflow. A vision-language model can interpret what it sees, yet perception alone does not make an action reliable. Microsoft’s UFO project is interesting because its evolution follows that boundary instead of pretending it does not exist. ...

May 9, 2025 · 15 min · 3008 words · Xinwei Xiong
Blank cards passing through a wooden ranking staircase toward a reader

LLM Recommendation Systems: Retrieval, Ranking, RAG, and Evaluation

Recommendation systems create a tempting illusion: the newer the model, the more advanced the system. Anyone who has operated one knows that the model is only the part above water. Below it sit exposure bias, latency budgets, feature freshness, inventory constraints, exploration risk, and counterfactual evaluation. Large language models add a useful semantic layer, but they do not make those older problems disappear. They also introduce new ones: higher cost, variable output, and explanations that can sound persuasive without reflecting the reasons an item was ranked. ...

April 23, 2025 · 18 min · 3653 words · Xinwei Xiong
A research notebook connecting source pages, citations, and synthesized outputs

Gemini Notebook (Formerly NotebookLM): A Source-Grounded Research Workflow

On July 16, 2026, Google renamed NotebookLM to Gemini Notebook. The old name will remain in search results, screenshots, and habits for a while, but the product did not disappear. Google describes it as the same standalone research tool, now connected more closely to the Gemini app and, eventually, Google Search. The rename also marks a larger change: notebooks are becoming places where software can not only read sources, but also run analysis against them. ...

April 21, 2025 · 18 min · 3787 words · Xinwei Xiong
A red, green, and blue testing loop surrounding an AI system

Test-Driven Development for AI and LLM Applications: A Practical 2026 Guide

Test-driven development is easy to explain when the function under test adds two numbers. Write a failing test, make it pass, then improve the implementation without changing its behavior. The difficulty begins when the function calls a language model and five different answers may all be acceptable. That does not make TDD obsolete. It means the word test has to become more precise. For an AI application, I use TDD as a discipline for discovering contracts. Some contracts are exact: a parser must reject malformed JSON, an authorization check must not leak another tenant’s data, and a tool call must match its schema. Other contracts are statistical: a support assistant should resolve most routine cases, cite the supplied policy, and rarely invent a refund rule. The first group belongs in ordinary tests. The second belongs in evaluations and monitoring. ...

April 21, 2025 · 16 min · 3402 words · Xinwei Xiong
Documents flowing through a measured conversion pipeline into structured Markdown

Microsoft MarkItDown 0.1.6: A Practical Document-to-Markdown Guide

A document converter is a bridge, not a source of truth. The important question is not whether the output looks clean at first glance, but whether the bridge preserves the evidence your next system needs. MarkItDown is easy to demonstrate: install a package, pass it a file, receive Markdown. The difficult work begins one minute later. A PDF may contain selectable text, scanned pages, diagrams, tables, or all four. A slide deck may hide essential numbers inside screenshots. A spreadsheet may be understandable only through formulas, merged cells, and spatial relationships. No single conversion mode handles every case equally well. ...

April 21, 2025 · 15 min · 3073 words · Xinwei Xiong
A checkpointed LangGraph StateGraph with recovery paths

LangGraph Architecture in 2026: StateGraph, Persistence, and Recovery

This project is an ongoing journey — learning AI open source projects with steady, daily progress. Through hands-on work with real projects and AI tooling, the goal is to develop the ability to solve complex problems and document the process. Notion List Basic Information: Project Name: LangGraph GitHub URL: langchain-ai/langgraph Main Tech Stack: Python, JavaScript/TypeScript, LangChain, LangSmith, checkpoint stores, LLM providers 1. What LangGraph Actually Solves LangGraph is the low-level orchestration framework and runtime in the LangChain ecosystem. It is useful when an agent must retain typed state, branch or loop, pause for a person, survive a process failure, or expose each transition to tracing and tests. You can use LangGraph without using LangChain’s model abstractions. ...

April 19, 2025 · 12 min · 2383 words · Xinwei Xiong
A restrained editorial illustration of connected paths representing models, agents, and graph workflows

LangChain 1.x in Production: Choosing Models, Agents, and LangGraph

LangChain used to be introduced as a box of chains, memory classes, prompt templates, loaders, and integrations. That description is historically accurate and operationally unhelpful. As of July 2026, the useful way to understand LangChain 1.x is much narrower: Call a model directly when the task is one inference with a clear input and output. Use create_agent when a model must choose among tools in a conventional model–tool loop. Use LangGraph when your application has durable state, explicit branches, retries, parallel work, long-running steps, or human decisions that are part of the workflow. This is not merely a taxonomy. It is an engineering rule about where complexity should live. Every framework abstraction creates a second system beside your product: another execution model, another vocabulary, and another place for failure to hide. The abstraction earns its place only when it makes the real system easier to operate. ...

April 16, 2025 · 15 min · 3178 words · Xinwei Xiong
Several model pipelines converge at an AI Gateway where cost and governance are weighed

AI Gateway Guide: LiteLLM, Kong, APISIX, Cloudflare, or Portkey?

An AI Gateway is not merely a reverse proxy placed in front of a language model. Once an application reaches production, every model call carries several kinds of uncertainty at once: long-lived streaming connections, token-based billing, provider quotas, sensitive inputs, changing model behavior, and outputs that cannot be trusted by default. A conventional API gateway can handle part of this traffic. It may authenticate clients, terminate TLS, enforce request limits, and route HTTP requests. It does not automatically answer the questions that matter most to an AI product: ...

April 16, 2025 · 16 min · 3223 words · Xinwei Xiong
GPT Researcher pipeline from query planning and retrieval to a cited report

GPT Researcher Guide: Python, Docker, MCP, Costs & Limits

A long report can look like certainty while merely arranging uncertainty more elegantly. The useful question is not how many pages an agent writes, but how a claim entered the report and whether a reader can walk back to its source. The short verdict GPT Researcher is a good fit when a team needs a programmable research pipeline, source traceability, and deployment control. It is not a truth machine. It automates planning, retrieval, context assembly, and report writing; it does not make weak pages authoritative or make every citation support the sentence beside it. ...

April 14, 2025 · 10 min · 2024 words · Xinwei Xiong
A layered search system connecting multimodal data, embeddings, reranking, and grounded answers

Jina AI in 2026: Embeddings, Reranking, Reader, and Jina Serve

Jina used to be easiest to explain as a cloud-native neural-search framework. That description is still true, but it is no longer sufficient. In 2026, the name covers two related products: Jina Search Foundation: hosted APIs and model families for embeddings, reranking, web reading, search, and research. Jina Serve: the open-source framework for turning Python AI components into services and composing them into distributed flows. The distinction matters. Search Foundation gives an application retrieval intelligence; Jina Serve gives a team control over how its own services run. One is a set of capabilities, the other an orchestration layer. Treating them as interchangeable usually leads either to unnecessary infrastructure or to a hosted dependency that was never consciously chosen. ...

April 12, 2025 · 12 min · 2381 words · Xinwei Xiong
LangChain 1.x architecture connecting models, agents, memory, retrieval, and observability

LangChain 1.x in Practice: A Developer Guide to Reliable Agents

A framework is most dangerous not when it lacks features, but when its old tutorials still look plausible. Familiar code can cross a version boundary long before it produces an obvious error. This guide follows the current LangChain Python 1.x path: model calls, agents, tools, structured output, middleware, memory, RAG, SQL, Ollama, and LangSmith. It is deliberately different from a framework overview. The aim is to leave you with a compact application architecture, runnable examples, and a way to recognize code that belongs to the 0.x era. ...

May 22, 2024 · 11 min · 2221 words · Xinwei Xiong
From Language Models to RAG: Capabilities, Limits, and Engineering

From Language Models to RAG: Capabilities, Limits, and Engineering

Introduction: Do Not Learn the Model from the Chat Window The first encounter with a large language model creates a powerful illusion. Something on the other side of the screen appears to have read widely, reasoned carefully, and chosen to explain itself. It can write code, summarize papers, preserve a tone across a conversation, and offer a polished rationale for an answer that is completely wrong. The chat window shows behavior, not mechanism. Fluency hides missing evidence. Completeness hides uncertainty. A human voice invites us to overestimate how firmly the model is connected to the world. ...

May 15, 2024 · 20 min · 4118 words · Xinwei Xiong
An abandoned film strip crossing a bridge toward a new modular video workflow

SoraEase After Sora: A Prompt Archive and Developer Migration Guide

In March 2024, this page was a long collection of Sora prompts. It belonged to the brief season when a research preview could become a small open-source movement before most people had touched the product. We copied examples, named camera movements, and tried to infer a grammar from a handful of remarkable clips. That guide can no longer honestly call itself a guide to “mastering Sora.” The Sora web and app experiences closed on April 26, 2026. OpenAI says its Sora API will be discontinued on September 24, 2026. A page that still teaches Sora as a growing platform would turn search traffic into misinformation. ...

March 14, 2024 · 13 min · 2722 words · Xinwei Xiong
A timeline of Sora from research preview and product launch to shutdown

Sora Retrospective: From Research Preview and Sora 2 to Shutdown

Status update, July 2026: The Sora website and app shut down on April 26, 2026. The Sora API is scheduled to shut down on September 24, 2026. This is no longer a guide to getting started; it is a record of what the technology, the product, and their ending can teach us. When I first wrote about Sora in February 2024, the irresistible detail was the one-minute video. Two years later, the more useful story is about boundaries: a research result is not a product specification, a better model does not guarantee a permanent service, safeguards do not erase risk, and generated media does not arrive with a simple answer to copyright. ...

February 24, 2024 · 10 min · 1942 words · Xinwei Xiong
Vector database learning

Vector Database Learning

Vector database learning I learned rust today and found that rust is a very suitable language for writing databases. I found a treasure project tikv. Its github project address is https://github.com/tikv/tikv . It is a very active project, but today I The topic is vector database in the field of AI. Without further ado, let’s get started right away. prerequisites 0.1 Introduction to basic knowledge: Understand the basic definition of vector database and its difference from traditional database. 0.2 Data Structure Basics: Learn vectors and other basic data structures, and how to represent and manipulate these structures in databases. 0.3 Introduction to Linear Algebra: Introduction to vector operations, including vector addition, subtraction and dot multiplication. 0.4 Similarity measures: Learn how to calculate similarity between vectors, such as cosine similarity. 0.5 Basics of Database Index: Introduces the basic concepts of database index, especially its application in vector databases. 0.6 Preliminary Search Algorithms: Learn basic search algorithms and understand how to perform effective searches in large data sets. 0.7 Application case studies: Study the application of vector databases in different fields (such as recommendation systems, image recognition). ...

January 20, 2024 · 41 min · 8650 words · Xinwei Xiong
From 2024 AI field notes to a reliable production RAG system

2024 AI Field Notes, Reassessed: From Emergence to Production RAG

This article began as notes from a large-language-model meetup in January 2024. Back then, the question was what a model could do. Two years later, the harder question is why anyone should trust the system around it. I have not polished the old notes into a story in which every early judgment was right. Mistakes are useful sediment. They remind us that a technical opinion is not a prophecy; it is a bet with an expiry date. Each section therefore keeps the 2024 observation and adds a 2026 reassessment: still true, changed, or wrong at the time. ...

January 14, 2024 · 11 min · 2234 words · Xinwei Xiong
A clockwork cart passing through controlled gates, representing the evolution from AutoGPT Classic to a governed agent platform

AutoGPT in 2026: From the Classic Experiment to the Platform

Status note, verified July 31, 2026: this article originally explained how to install the 2023 stand-alone Auto-GPT agent. Those commands are obsolete. The official project now says that AutoGPT Classic is unsupported, its dependencies will not be updated, and it has known security issues. Treat Classic as a historical laboratory, not a production tool. For new work, use the maintained AutoGPT Platform or another actively maintained workflow system. In the spring of 2023, Auto-GPT made a compelling promise: describe an objective, give a model some tools, and watch it plan its own way forward. I installed it because the idea felt less like a chatbot feature and more like a change in the shape of software. ...

March 18, 2023 · 10 min · 2100 words · Xinwei Xiong