Terms
Understand the concepts and technical vocabulary in our conversations.
178 termsTerms
2D
Also known as二维图形 · 二维
2D refers to a plane with only two dimensions, length and width. 2D computer graphics work directly with pixels or vector shapes on a flat surface rather than building solid models, common in illustration, interface icons and animation.
Term details→023D
Also known as三维图形 · 三维
3D refers to space with three dimensions—length, width and height. In computer graphics, 3D describes techniques that model solid objects mathematically and render them into planar images, widely used in film, games and industrial design.
Term details→034K resolution
Also known as4K
4K describes image resolutions around four thousand pixels wide: consumer UHD commonly uses 3840×2160, while digital cinema also uses 4096×2160.
Term details→045G
Also known as第五代移动通信技术 · 5G NR · 第五代移动通信
5G is the fifth generation of mobile network technology, standardized by 3GPP to deliver higher data rates, lower latency, and greater connection density than 4G. It introduces a new radio interface (5G NR) and a more flexible core network, supporting both enhanced mobile broadband and internet-of-things use cases.
Term details→05A-Roll and B-Roll
Also known asA-Roll · B-Roll · 主体素材 · 补充素材
A-roll carries the main narrative, while B-roll supplies supporting or illustrative footage. The labels describe editorial roles, not image quality.
Term details→06A/B Testing
Also known asA/B Test · AB 测试 · Split Testing
A/B testing is a controlled experiment method: users or traffic are randomly split into groups that experience different variants, and a metric chosen in advance decides which variant performs better.
Term details→07Agent
Also known asAI Agent · AI 智能体
A software system that perceives its environment, makes decisions and uses tools to pursue goals, often with an LLM serving as its reasoning core.
Term details→08Agent Harness
Also known asHarness
The runtime layer around a model that provides tool access, context management, loop control and result verification, turning the model into an agent that can actually execute tasks.
Term details→09Agent Memory
Also known asMemory
Agent memory is the mechanism by which an AI system persists and reuses information across sessions, so the model still knows a user's facts, preferences, and ongoing work after a conversation ends. In the agent context it often lives in files or databases, distinct from the context of a single conversation.
Term details→10Agent Skills
Also known asSkill · Skills · Agent Skill
A mechanism that packages instructions, scripts and resources into folders (such as a directory containing a SKILL.md file) so an agent can load the right capability on demand.
Term details→11Agent steering
Also known asSteer · 引导模式
Agent steering supplies additional instructions or corrections while an agent is working so subsequent actions can respond to new input.
Term details→12Agent Teams
Also known asMulti-Agent System · MAS · 多智能体系统
An arrangement in which multiple agents divide work, communicate and coordinate to complete complex tasks that a single agent would struggle with alone.
Term details→13Aha Moment
Also known as顿悟时刻 · 啊哈时刻
An aha moment is a sudden realization or experience of a product’s value: the point at which its usefulness becomes clear.
Term details→14AI Native
Also known asAI-Native · AI 原生应用
Describes a product, company or workflow designed from the ground up with AI as a core capability, rather than bolting AI onto existing systems afterward.
Term details→15AI Slop
Also known asSlop
Low-quality digital content mass-produced with generative AI, typically created at volume to gain traffic, earn money or deceive people.
Term details→16Annual Recurring Revenue (ARR)
Also known asARR · 年化经常性收入 · 年度经常性收入 · Annual Recurring Revenue
Annual recurring revenue (ARR) is the annualized value of the recurring revenue a subscription business currently books from active contracts — the run rate that income would reach if it stayed constant for a year. It counts only recurring contract revenue and normally excludes one-time charges.
Term details→17API key
Also known asAPI 密钥
An API key is an identifier or credential supplied with API requests, often linking usage to a project, quota, or billing account.
Term details→18Application Programming Interface (API)
Also known asAPI · Web API · 接口 · 应用程序编程接口 · Application Programming Interface
An application programming interface (API) is a contract that specifies how software components call each other's functions and exchange data. Callers follow the documented methods, parameters and response formats without needing to know the internals behind them.
Term details→19Application Server
Also known asApp Server
An application server is a software environment that hosts an application's business logic and serves application-level requests to clients, such as Tomcat or a Node.js runtime, typically handling dynamic requests behind a web server.
Term details→20Artificial General Intelligence (AGI)
Also known asAGI · 通用人工智能 · Artificial General Intelligence
Artificial general intelligence is a hypothetical AI system that matches human performance across a broad range of cognitive tasks and transfers its skills to unfamiliar domains. It remains a research goal with no widely acknowledged implementation.
Term details→21Artificial Intelligence (AI)
Also known asAI · 人工智能 · Artificial Intelligence
Artificial intelligence is an umbrella term for computer systems that perform tasks typically associated with human intelligence, such as learning, reasoning, perception, and language processing. It names a whole field of techniques, not one specific algorithm or product.
Term details→22Artificial Superintelligence (ASI)
Also known asASI · 人工超级智能 · Artificial Superintelligence
Artificial superintelligence is a hypothetical AI system whose intelligence greatly exceeds human intelligence in virtually every domain. It exists only in theory and forecasting discussions, with no accepted path to building it.
Term details→23Attention-Deficit/Hyperactivity Disorder (ADHD)
Also known asADHD · 注意缺陷多动障碍 · Attention-Deficit/Hyperactivity Disorder
Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder marked by persistent inattention, hyperactivity and impulsivity that usually begins in childhood and can continue into adulthood. Diagnosis and treatment belong to qualified professionals working from standardized criteria.
Term details→24Augmented Reality (AR)
Also known asAR · 增强现实 · Augmented Reality
Augmented Reality (AR) is a technology that overlays virtual content on the user's view of the real environment, with the overlay interactive in real time and aligned to the physical scene. Phone camera effects and see-through glasses are common AR carriers.
Term details→25Automatic Speech Recognition (ASR)
Also known asASR · Speech-to-Text · 语音识别 · 自动语音识别 · Automatic Speech Recognition
Automatic speech recognition (ASR) is the technology that transcribes spoken audio into text automatically, also known as speech-to-text. It powers captioning, voice input and meeting transcription.
Term details→26Beats per Minute (BPM)
Also known asBPM · 每分钟节拍数 · Beats per Minute
Beats per minute (BPM) is a unit for measuring musical tempo—the number of beats occurring in one minute; higher values mean a faster pace. It is the standard way to quantify tempo and appears throughout music production and playback software.
Term details→27Benchmark
A benchmark measures a system's performance using a fixed dataset and a uniform scoring rule. In AI, it is typically a set of tasks plus automatic scoring, used to grade model capability and compare models against each other.
Term details→28Bill of Materials (BOM)
Also known asBOM · 物料清单 · Bill of Materials
A bill of materials (BOM) is a structured list of all parts, materials and quantities required to manufacture a product, often organized hierarchically to describe how components fit together; it underpins purchasing, inventory and production planning.
Term details→29Biometrics
Biometrics uses measurable physical or behavioral characteristics to identify a person or verify a claimed identity.
Term details→30Bluetooth
Bluetooth is a family of wireless connectivity technologies used for audio, device data exchange, and other communication scenarios.
Term details→31Browser Use
Also known asbrowser automation
Browser use is the ability of AI agents to complete tasks through a web browser: reading pages, clicking, filling forms and navigating. Browser Use is also the name of an open-source project that connects browser automation to LLM agents.
Term details→32Bug
Also known as程序错误 · 缺陷
A bug is a defect in a program that causes behavior to deviate from what is expected, produce errors or crash. The word was already in engineering use, but the 1947 record of a moth found in the Harvard Mark II made it widespread.
Term details→33Business-to-Business (B2B)
Also known asB 端 · To B · B2B · 面向企业 · Business-to-Business
Business-to-business (B2B) describes companies that sell products or services to other companies or organizations rather than to individual consumers — think enterprise software, cloud services and industrial goods.
Term details→34Business-to-Consumer (B2C)
Also known asC 端 · To C · B2C · 面向消费者 · Business-to-Consumer
Business-to-consumer (B2C) describes companies that sell products or services directly to individual consumers, as in e-commerce, streaming subscriptions and retail chains.
Term details→35Cache
A cache stores reusable data or computed results so later requests can avoid repeating the full retrieval or computation.
Term details→36Callback
Also known as回调
A callback is a function passed to other code so that the receiving code can invoke it at an appropriate point.
Term details→37Central Processing Unit (CPU)
Also known asCPU · 处理器 · processor · 中央处理器 · Central Processing Unit
A central processing unit (CPU) is the core chip that executes program instructions and carries out the computation and control of a computer. It processes instructions through a fetch-decode-execute cycle, with performance shaped by core count, clock speed, and architecture design.
Term details→38Chain of thought (CoT)
Also known as思维链 · CoT · Chain of Thought
A chain of thought is a sequence of intermediate reasoning steps generated by a model; related prompting methods use such steps for multi-step problems.
Term details→39Chatbot
Also known asChat Bot
A chatbot is a program that interacts with people through a conversational interface, mainly in the form of text Q&A, evolving from rule-based scripts to large-language-model-driven systems. It is usually built around a single turn or session of dialogue, which is also the common dividing line from agents.
Term details→40Chief Executive Officer (CEO)
Also known asCEO · 首席执行官 · Chief Executive Officer
The chief executive officer (CEO) is a company's top-ranking operating manager, responsible for setting and executing strategy, appointed by and answerable to the board of directors. Other C-suite officers (CTO, CFO and so on) typically report to the CEO.
Term details→41Chief Technology Officer (CTO)
Also known asCTO · 首席技术官 · Chief Technology Officer
The chief technology officer (CTO) is the executive in charge of a company's technology strategy and engineering organization, setting the technical direction and leading the R&D team that builds the product, usually reporting to the CEO.
Term details→42Cloud Computing
Also known asCloud · 云服务
Cloud computing, as defined by NIST SP 800-145, is a model for on-demand, self-service access to configurable computing resources (networks, servers, storage, applications and services) that can be rapidly provisioned and released with minimal management effort, usually billed by usage.
Term details→43Coding Agent
Also known as编程智能体 · 编码智能体
An agent for software development that can inspect a repository, edit files, run commands and tests, and continue working from the results.
Term details→44Command-Line Interface (CLI)
Also known asCLI · 命令行 · command line · 命令行界面 · Command-Line Interface
A command-line interface (CLI) is a way to interact with programs through text commands: the user types commands and arguments in a terminal, and the program responds with text output. It is the foundation of scripting and automation.
Term details→45Computer Use
Also known ascomputer use agent · CUA
Computer use is the ability of a model to operate a computer's graphical interface the way a person does — reading the screen, moving the cursor, clicking and typing — to complete tasks across applications. Anthropic and OpenAI both ship computer-use tools or model capabilities.
Term details→46Computer Vision (CV)
Also known asCV · 计算机视觉 · Computer Vision
Computer vision extracts and interprets information from images or video, supporting tasks such as recognition, detection and segmentation.
Term details→47Concurrency
Concurrency organizes multiple tasks so they can make progress over overlapping periods of time.
Term details→48Connector
Also known asIntegration
A configured integration that links an agent or AI assistant to external tools, data sources and services; in Claude, connectors are largely built on MCP and can be installed from the official directory.
Term details→49Context
Context is everything a language model can see while handling one request — the user's input, conversation history, and attached documents or tool results. Its capacity limit, measured in tokens, is called the context window.
Term details→50Cookie
Also known asHTTP Cookie
A cookie is a small piece of key-value data that a website stores in the browser via the Set-Cookie response header; the browser sends it back with subsequent requests, enabling sessions, personalization and state. It is defined by RFC 6265.
Term details→51Database
A database is an organized collection of stored data, usually accessed and updated through a database management system.
Term details→52Deep research
Also known as深度研究
In AI applications, deep research involves multiple rounds of searching, reading, analysis, and synthesis rather than a single search result.
Term details→53Diffusion Model
Also known asDiffusion · 去噪扩散模型
A diffusion model is a generative model trained to reverse a gradual noising process: at generation time it starts from pure noise and denoises step by step to produce images, audio and other samples. It is known for high output quality and stable training, at the cost of multi-step inference.
Term details→54Digital certificate
Also known as根证书
A digital certificate associates a public key with information about a subject, supporting authentication through signatures and a chain of trust.
Term details→55Do It Yourself (DIY)
Also known as自己动手做 · Do It Yourself
Do it yourself (DIY) refers to building, repairing or modifying things on your own instead of hiring professionals, with the emphasis on doing it yourself rather than on professional-grade results.
Term details→56Document Object Model (DOM)
Also known asDOM · 文档对象模型 · Document Object Model
The Document Object Model (DOM) is the tree of objects a browser builds when parsing an HTML or XML document, with nodes representing elements, text and attributes. Programs read and write this tree through standard interfaces to inspect and change the page dynamically.
Term details→57Dots per Inch (DPI)
Also known asDPI · 每英寸点数
DPI stands for dots per inch. For a mouse, it commonly describes sensitivity: a higher setting produces more pointer movement for the same physical travel.
Term details→58Embodied AI
Also known asEmbodied Intelligence · 具身人工智能
Embodied AI refers to AI agents that perceive, act and learn through a physical body interacting directly with their environment, stressing that intelligence emerges from a perceive-decide-act loop rather than disembodied computation. Robots are its most typical carrier.
Term details→59Engagement Rate
Engagement rate measures how much an audience interacts with content, usually computed as the number of interactions such as likes, comments and shares divided by impressions, reach or followers, and expressed as a percentage.
Term details→60Environment variable
Environment variables are named values supplied to a process by its execution environment, commonly for configuration, paths, or service connections.
Term details→61eSIM
Also known asembedded SIM · 嵌入式SIM卡 · 嵌入式 SIM
eSIM (embedded SIM) is a form of SIM that is soldered directly into a device, with carrier profiles downloaded and activated remotely instead of inserting a physical card. Defined by GSMA remote SIM provisioning specifications, it has been widely adopted in phones, watches, and IoT devices.
Term details→62Evaluation
Also known aseval · evals · LLM evaluation
In AI, evaluation (often shortened to eval) is the process of measuring model output quality and task performance using test data, metrics, and human or automated judging. It underpins model selection, version comparison and regression testing.
Term details→63Everyday Carry (EDC)
Also known asEDC · Everyday Carry
EDC refers to useful items a person routinely carries, and to the practice of selecting that everyday kit.
Term details→64FAANG
FAANG groups Facebook, Amazon, Apple, Netflix, and Google using their initial letters.
Term details→65Firmware
Firmware is program code and data held in a device’s nonvolatile storage to implement basic device functions.
Term details→66Fork
In software development, a fork is an independent copy of an existing project or repository that can be modified and evolved on its own. On platforms like GitHub, a fork is a personal copy of someone else's repository, and changes can be contributed back through pull requests.
Term details→67Frame generation
Frame generation synthesizes additional images to increase displayed frame rate and perceived motion smoothness.
Term details→68Frame rate (FPS)
Also known as帧率 · FPS
Frame rate is the number of video or animation frames displayed or processed per second, commonly expressed as FPS.
Term details→69Galgame
Also known as美少女游戏 · Bishōjo game · Gal Game
Galgame commonly refers to games centered on interactions with female characters, often using dialogue, relationships and branching stories.
Term details→70Game production tiers (AAA / AA)
Also known as3A · 2A · AAA · AA
AAA and AA are informal game-industry labels for production investment, team scale, and content scope.
Term details→71Generative Engine Optimization (GEO)
Also known asGEO · 生成式引擎优化 · Generative Engine Optimization
Generative engine optimization (GEO) is an emerging practice of organizing and adjusting content so that generative engines are more likely to retrieve, cite and surface it in AI-generated answers. The concept spread from the research paper GEO: Generative Engine Optimization, published in November 2023.
Term details→72GIF
Also known asGraphics Interchange Format
GIF (Graphics Interchange Format) is a bitmap image format released by CompuServe in 1987, using lossless LZW compression, at most 256 colors per frame, with support for frame-by-frame animation. The version in common use is defined by the 1989 GIF89a specification.
Term details→73GNU General Public License (GPL)
Also known asGPL · GPLv2 · GPLv3 · GNU 通用公共许可证 · GNU General Public License
The GPL is a copyleft license published by the Free Software Foundation: anyone may use, modify and redistribute the software, but derivative works must be distributed under the same license with source code available. Common versions are GPLv2 and GPLv3.
Term details→74Graphical User Interface (GUI)
Also known asGUI · 图形界面 · 图形用户界面 · Graphical User Interface
A graphical user interface (GUI) lets people operate a computer through visual elements such as windows, icons, menus and a pointer, replacing text-only command lines that had to be memorized.
Term details→75Graphics Processing Unit (GPU)
Also known asGPU · 显卡 · graphics card · 图形处理器 · Graphics Processing Unit
A graphics processing unit (GPU) is a processor designed for massively parallel computation, originally built to accelerate graphics rendering and now widely used for general-purpose parallel workloads such as AI training and inference. It combines thousands of lightweight cores that excel at running many similar arithmetic tasks at once.
Term details→76Gross Domestic Product (GDP)
Also known asGDP · 国内生产总值 · Gross Domestic Product
Gross domestic product (GDP) measures the market value of all final goods and services produced by an economy in a given period, usually a quarter or a year. It is the most widely used aggregate gauge of macroeconomic size and growth.
Term details→77High-Definition Multimedia Interface (HDMI)
Also known asHDMI · 高清多媒体接口
HDMI is an interface for carrying digital video and audio between devices such as computers, televisions and displays.
Term details→78Hook
Also known as钩子函数
A hook is an extension point a program reserves at specific events or lifecycle stages, letting external code inject custom logic without modifying the main program. The mechanism is widely used in Git, front-end frameworks and AI agent toolchains.
Term details→79Human in the Loop (HITL)
Also known asHITL · 人机协同 · 人在回路 · Human in the Loop
Human in the Loop (HITL) is a mechanism that keeps a person involved in an automated or AI workflow: the machine executes tasks while a human reviews, decides or corrects at key checkpoints. Human judgment is embedded in the automated process rather than replacing it.
Term details→80Human Resources (HR)
Also known asHR · 人力资源 · Human Resources
Human resources (HR) is the organizational function responsible for hiring, training, compensation, performance and employee relations; the term also refers to the department and the people who work in it. Its remit spans the whole life cycle of employees in the organization.
Term details→81HyperText Markup Language (HTML)
Also known asHTML · 超文本标记语言 · HyperText Markup Language
HyperText Markup Language (HTML) is the markup language that defines the structure and content of web pages, using elements and attributes to describe headings, paragraphs, links, images and more. It is maintained by WHATWG as the HTML Living Standard and underpins every web page.
Term details→82Infrastructure
Also known asInfra · IT 基础设施
Infrastructure is the combined foundation of hardware, software and services that supports applications and business operations, including compute, storage, networking, facilities and the surrounding operational practices. It is often shortened to infra.
Term details→83Initial Public Offering (IPO)
Also known asIPO · 首次公开募股
An IPO generally refers to a company’s first sale of shares to the public.
Term details→84Integrated Development Environment (IDE)
Also known asIDE · 开发环境 · 集成开发环境 · Integrated Development Environment
An integrated development environment (IDE) is software that brings code editing, building, debugging and running together in one graphical application. Well-known examples include Xcode, Android Studio and Visual Studio Code.
Term details→85Internet
Also known as因特网
The internet is a global system of interconnected computer networks that communicate using the TCP/IP protocol suite, carrying services such as the web and email.
Term details→86Internet Data Center (IDC)
Also known asIDC · 数据中心 · 互联网数据中心 · Internet Data Center
An IDC (Internet Data Center) is a facility that centrally houses servers, storage and network equipment with reliable power, cooling, bandwidth and operations support; it is the physical foundation of internet services and cloud computing.
Term details→87Jevons Paradox
The Jevons paradox is an observation in economics: when technological progress makes a resource more efficient and cheaper per unit, total consumption of that resource often rises rather than falls. William Stanley Jevons raised it in 1865 with the example of more efficient steam engines increasing coal use; today it is often applied to AI compute and to demand expanding as model costs fall.
Term details→88JSON
Also known asJavaScript Object Notation
JSON (JavaScript Object Notation) is a text-based data interchange format defined by RFC 8259, expressing structured data with objects, arrays, strings, numbers, booleans and null. It is one of the most common formats for web APIs and configuration files.
Term details→89Key Opinion Leader (KOL)
Also known asKOL · 关键意见领袖 · Key Opinion Leader
A key opinion leader (KOL) is a person with recognized expertise in a specific field whose views materially influence the attitudes and decisions of a target audience. The term is common in marketing and covers experts, academics and creators.
Term details→90Key Performance Indicator (KPI)
Also known asKPI · 关键绩效指标 · Key Performance Indicator
A key performance indicator (KPI) is a quantifiable measure an organization selects to gauge progress toward an important objective, such as monthly active users, customer churn rate or gross margin. Good KPIs tie directly to strategy and are few in number.
Term details→91Know Your Customer (KYC)
Also known asKYC · 客户身份核验
KYC refers to processes for understanding and verifying a customer’s identity and background, commonly within financial customer due diligence.
Term details→92KV Cache
Also known as键值缓存
The KV cache stores the key and value vectors of attention layers during transformer inference, so generating each new token does not require recomputing attention over earlier tokens. It is a key optimization for generation speed.
Term details→93Large Language Model (LLM)
Also known asLLM · 大语言模型 · Large Language Model
A large language model is a deep neural network trained on massive amounts of text to predict sequences of tokens. It handles a broad range of language tasks, from generation and question answering to translation and coding.
Term details→94Limited Liability Company (LLC)
Also known asLLC · 有限责任公司 · Limited Liability Company
A limited liability company (LLC) is a business structure, mainly used in the United States, that keeps member liability limited to their investment while leaving room for flexible taxation and internal management.
Term details→95Linux distribution
A Linux distribution combines the Linux kernel with user programs, system tools, and software management into an installable operating system.
Term details→96llms.txt
llms.txt is a proposed file convention for giving language models and agents a concise information entry point to a website.
Term details→97Local area network (LAN)
Also known as局域网 · LAN
A local area network connects devices within a limited area, such as a home, office, or building.
Term details→98Low-Rank Adaptation (LoRA)
Also known asLoRA · 低秩适应 · Low-Rank Adaptation
LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning method: the model's original weights stay frozen, and only small low-rank matrices injected alongside selected layers are trained, adapting the model to a task or style with a fraction of the trainable parameters. The resulting artifacts are small and easy to distribute and merge.
Term details→99Markdown
Also known asmd
Markdown is a lightweight markup language that uses plain-text symbols (such as
Term details→100Mod
Also known asMods · 模组
A mod is a player- or third-party-made modification of a video game, adjusting its content, mechanics or presentation, ranging from small item tweaks to full overhauls. Both the practice of modifying and its output are commonly called mods.
Term details→101Model Context Protocol (MCP)
Also known asMCP · 模型上下文协议 · Model Context Protocol
An open protocol that gives AI applications a standardized interface for connecting to external data sources and tools, replacing fragmented point-to-point integrations.
Term details→102Model distillation
Also known as知识蒸馏
Model distillation trains a student model using a teacher’s outputs or behavior to transfer some of the teacher’s capabilities.
Term details→103Model inference
Model inference applies a trained model to inputs to produce predictions or generated outputs.
Term details→104Musical Instrument Digital Interface (MIDI)
Also known asMIDI · 乐器数字接口 · Musical Instrument Digital Interface
Musical Instrument Digital Interface (MIDI) is a protocol for electronic instruments, computers and other devices to exchange performance data such as notes, velocity and control messages, released in 1983 and maintained by the MIDI Association. MIDI carries instructions, not audio itself; the sound is produced by the receiving device.
Term details→105NAND Flash
Also known asNAND · NAND flash memory
NAND flash is a non-volatile storage technology that retains data without power, named for the series wiring of its memory cells that resembles a NAND gate. It stores bits as electrical charge in floating-gate or similar cell structures and serves as the underlying media for SSDs, memory cards, and USB drives.
Term details→106Neocloud
Also known asNeo-cloud · 新型算力云
Neocloud is a term popularized by the industry research firm SemiAnalysis in 2024 reports for a new generation of cloud providers focused on GPU capacity for AI training and inference; compared with hyperscale clouds, they emphasize bare-metal and GPU cluster rentals with lean operations.
Term details→107Network
A network is a set of computers and devices connected by communication links and protocols to share resources and exchange data, classified by reach into LANs, MANs and WANs.
Term details→108Network-Attached Storage (NAS)
Also known asNAS · 网络存储 · 网络附加存储 · Network-Attached Storage
Network-attached storage (NAS) is a dedicated device that provides file-level storage services to multiple clients over a standard network. Typically a standalone box with multiple drives running its own operating system, it lets a home or team store and share files centrally.
Term details→109One Shot
Also known asOne-Shot · 一次完成
In the show’s AI creation and coding discussions, one shot means obtaining a usable result from an initial request rather than through repeated follow-up revisions.
Term details→110One-Person Company (OPC)
Also known asOPC · 一人公司 · One-Person Company
A one-person company (OPC) is a company or way of working founded and run by a single person, usually with no or very few employees, with one person covering everything from product to growth using tools and outsourcing. In Weekly
Term details→111Open source
Open source describes software distributed under a license permitting use, inspection, modification, and redistribution of its source code.
Term details→112Operating System (OS)
Also known asOS · 操作系统 · Operating System
An operating system (OS) is system software that manages a computer's hardware resources and provides the environment in which applications run, handling process scheduling, memory management, file systems and device drivers. Windows, macOS and Linux are common examples.
Term details→113Personal Agent
Also known as个人智能体
An agent that serves a single individual, builds up knowledge of that person's preferences and context, and handles everyday tasks on their behalf.
Term details→114Personal Computer (PC)
Also known asPC · 个人电脑 · 个人计算机 · Personal Computer
A personal computer (PC) is a computer designed for use by one person at a time, typically built around a processor, memory, a display and a keyboard. In a narrower sense, "PC" often refers specifically to IBM PC compatibles (mostly running Windows), contrasted with the Mac; broadly, both are personal computers.
Term details→115Plaintext password storage
Also known as明文保存
Plaintext password storage retains the original readable password instead of a password-verification hash.
Term details→116Platform as a service (PaaS)
Also known asPaaS
Platform as a service provides an environment for deploying and running applications while the provider manages underlying infrastructure.
Term details→117Plugin
Also known asPlug-in
In AI coding tools, an extension mechanism that bundles commands, subagents, MCP servers and hooks into a single installable unit; more generally, a component that adds features to a host program without modifying its core.
Term details→118Portable Document Format (PDF)
Also known asPDF · 便携式文档格式 · Portable Document Format
PDF is a fixed-layout document format introduced by Adobe in 1993 and now an ISO standard (ISO 32000). It packs text, fonts, images and layout into one file so the document renders consistently across devices and platforms.
Term details→119Product-Market Fit (PMF)
Also known asPMF · 产品市场契合 · 产品与市场契合 · Product-Market Fit
Product-market fit (PMF) is the state in which a product genuinely satisfies a real, sizable market need, so that demand pulls the product forward. The concept was popularized by investor Marc Andreessen.
Term details→120Prompt
Also known as提示
A prompt is the input given to a generative model — natural language, examples or constraints that state what task to perform and in what form to answer. The practice of writing effective prompts is called prompt engineering.
Term details→121Quality Assurance (QA)
Also known asQA · 质量保证 · Quality Assurance
QA (quality assurance) covers the activities and processes in software engineering meant to prevent defects and assure product quality; it also names the role or team that does this. In Weekly #001, QA appears in the job-role sense (listed alongside design, front-end, and back-end), referring to quality assurance rather than question answering (Q&A).
Term details→122RAID
Also known as磁盘阵列 · 冗余磁盘阵列 · redundant array of independent disks · 独立磁盘冗余阵列
RAID (redundant array of independent disks) is a technique that combines multiple physical drives into one logical storage unit, using data distribution and redundancy to improve performance, capacity, or fault tolerance. Common levels such as RAID 0, 1, 5, 6, and 10 make different trade-offs between speed, redundancy, and capacity.
Term details→123Recursive Self-Improvement (RSI)
Also known asRSI · 递归自我改进
RSI describes a loop in which a system improves itself and uses the resulting capability to support further improvement.
Term details→124Regression testing
Regression testing checks whether changes to software or its environment have unintentionally affected existing behavior.
Term details→125Regular expression
Also known as正则 · Regex
A regular expression describes a text-matching pattern for searching, extracting, replacing, or validating strings.
Term details→126Reinforcement Learning from Human Feedback (RLHF)
Also known asRLHF · 基于人类反馈的强化学习 · Reinforcement Learning from Human Feedback
Reinforcement learning from human feedback (RLHF) is a training method in which humans compare or rank a model's outputs, a reward model is trained from those judgments, and reinforcement learning then optimizes the model to produce outputs that better match human preferences.
Term details→127Release Notes
Also known aschangelog · 更新日志
Release notes are documents shipped with each software release telling users what was added, changed, fixed and still broken. Common forms include a changelog and GitHub Release pages.
Term details→128Rendering
Rendering computes final images from scenes, source media, and effects, producing still images or successive video frames.
Term details→129Return on Investment (ROI)
Also known asROI · 投资回报 · 投资回报率 · Return on Investment
Return on investment (ROI) is a ratio that expresses how much a given outlay earned back, usually computed as net gain divided by cost and quoted as a percentage. Its scope is flexible — it can be applied to a single ad campaign, a piece of equipment, or an entire project.
Term details→130Roguelike
Also known as类 Rogue 游戏
Roguelike is a video game genre descended from the 1980 game Rogue, commonly identified by turn-based play, grid-based maps, procedurally generated levels and permadeath. Each run offers different maps and encounters, and death means starting over.
Term details→131Root access
Also known asRoot
Root commonly denotes the superuser identity on Unix-like systems, with administrative privileges unavailable to ordinary users.
Term details→132RSS
Also known asReally Simple Syndication
RSS is an XML-based format for syndicating web content: a site publishes titles, summaries and links in a feed file that subscribers poll with a feed reader to pick up updates. The RSS 2.0 specification is maintained by the RSS Advisory Board.
Term details→133Runtime
Also known as运行时环境 · runtime environment
A runtime is the environment and supporting system in which a program actually executes, covering memory management, scheduling, standard libraries and interaction with the operating system. The word also refers to the concrete software that hosts a language's execution, such as Node.js or a browser's JavaScript engine.
Term details→134Saving time and passing time
Also known asSave Time · Kill Time
Save Time emphasizes reducing the time needed for a task; Kill Time usually means passing time. The episode uses the contrast to discuss utility and entertainment.
Term details→135Scalable Vector Graphics (SVG)
Also known asSVG · 可缩放矢量图形 · Scalable Vector Graphics
Scalable Vector Graphics (SVG) is an XML-based vector image format defined by the W3C, describing shapes, paths and text mathematically so images scale to any size without loss. Unlike pixel-based bitmap formats such as GIF, SVG is commonly used for icons and illustrations.
Term details→136Search Engine Optimization (SEO)
Also known asSEO · 搜索引擎优化 · Search Engine Optimization
Search engine optimization (SEO) is the practice of improving content, technical structure and backlinks so that pages get indexed and ranked better in the organic (non-paid) results of search engines.
Term details→137Secure Shell (SSH)
Also known asSSH · Secure Shell Protocol · 安全外壳协议 · Secure Shell
SSH is a protocol for securely operating network services over an unsecured network (RFC 4251), providing encryption and authentication. It is best known for remote login and command-line administration, and is also used for secure file transfer and port forwarding.
Term details→138Semantic Versioning (SemVer)
Also known asSemVer · 语义化版本号 · 语义化版本 · Semantic Versioning
Semantic Versioning (SemVer) is a versioning scheme with the format major.minor.patch (for example 2.4.1): a major bump signals an incompatible API change, a minor bump adds backward-compatible functionality, and a patch bump fixes backward-compatible bugs.
Term details→139Server
A server is a computer or program that provides services to other programs or devices (clients). In a hardware sense it refers to the computing equipment in data centers; in a software sense it is a process that listens for and answers requests, such as a web server.
Term details→140Session
Also known as会话期
A bounded period of continuous interaction between a user and a system, during which related state and context are preserved; in AI assistant contexts, one complete conversation and the context it carries.
Term details→141Show Notes
Also known as节目笔记
Show notes are text materials published alongside a podcast episode, typically covering key points, guest information, timestamps and related links that help listeners understand, search and revisit the episode.
Term details→142Smart home
A smart home uses connected devices, sensors, and control systems for remote operation, status feedback, or household automation.
Term details→143Software as a Service (SaaS)
Also known asSaaS · 软件即服务 · Software as a Service
Software as a service (SaaS) is a software delivery model in which applications are hosted by the vendor in the cloud and used over the network on a monthly or annual subscription, with no local installation or maintenance. It is one of the main service models of cloud computing.
Term details→144Software Development Kit (SDK)
Also known asSDK · devkit · 软件开发工具包 · Software Development Kit
A software development kit (SDK) is a bundle of resources for building on a specific platform or service, typically including libraries, APIs, documentation, sample code and build or debugging tools. It helps developers build apps or integrate services on a target platform faster.
Term details→145Solid-State Drive (SSD)
Also known asSSD · solid state drive · 固态硬盘 · Solid-State Drive
A solid-state drive (SSD) is a storage device that keeps data in NAND flash memory chips rather than on spinning magnetic platters, with a controller managing reads, writes, and wear. With no mechanical parts, it offers lower access latency and better shock resistance than a hard disk drive.
Term details→146State of the Art (SOTA)
Also known asSOTA · state-of-the-art · 最先进水平 · 当前最佳水平 · State of the Art
SOTA (state of the art) refers to the best publicly known result or the most advanced method on a given task. Saying a model "achieves SOTA" usually means it matches or beats the best publicly reported score on recognized benchmarks.
Term details→147Streaming output
Streaming output delivers portions of a result before the whole result has finished generating.
Term details→148Subagent
Also known asSub Agents · Sub-agent
An agent spawned by a main agent to carry out a specific subtask, usually running with its own context and tools and returning only a summary of results.
Term details→149System One
Also known asSystem One Model
System One is a model category proposed by TypeSafe AI, officially defined as "a new class of frontier models built to make fast, structured decisions that software can use directly"; the name borrows the System 1/System 2 distinction from Kahneman's Thinking, Fast and Slow. TypeSafe calls its Jev the first publicly announced System One model.
Term details→150System Prompt
Also known asSystem Message · 系统指令
A system prompt is an instruction injected by the developer or product before a conversation begins, setting the model's role, rules and output constraints; it is typically invisible to end users.
Term details→151Technology stack
A technology stack is the combination of languages, frameworks, databases, and infrastructure used to build and run an application.
Term details→152Tensor Processing Unit (TPU)
Also known asTPU · 张量处理器 · 张量处理单元 · Tensor Processing Unit
A tensor processing unit (TPU) is an application-specific integrated circuit (ASIC) designed by Google for machine learning workloads, built to accelerate tensor operations in neural networks. Centered on computations such as matrix multiplication, it is offered primarily as a service through Google Cloud.
Term details→153Text-to-Speech (TTS)
Also known asTTS · Speech Synthesis · 语音合成 · 文本转语音 · Text-to-Speech
Text-to-speech (TTS), also called speech synthesis, is the technology that converts written text into audible speech. Modern neural TTS can produce near-human voices and clone a voice from a short sample.
Term details→154Token
A token is the basic unit of text that a language model reads and generates. A tokenizer splits raw text into tokens, each corresponding to a word, subword, character, or symbol; model input, output, and usage are all measured in tokens.
Term details→155Token Maxxing
Token Maxxing is a phrase used by Next Token participants for pushing AI usage to the max and consuming tokens at massive scale. It came up in Weekly
Term details→156Tokens per Second (TPS)
Also known asTPS · 每秒词元数 · Tokens per Second
Tokens per second (TPS) measures how fast a language model generates text — the number of tokens it outputs per second, usually counted after the first token arrives. Higher values mean text appears faster.
Term details→157Transformer
The transformer is a neural network architecture built on self-attention, introduced in the paper "Attention Is All You Need" for processing sequences such as text; it is the architecture underlying large language models.
Term details→158Universal Basic Income (UBI)
Also known asUBI · 全民基本收入 · Universal Basic Income
Universal basic income (UBI) is a policy proposal under which a government or public institution pays every member of a community the same amount of cash on a regular, unconditional basis, with no means test and no work requirement.
Term details→159Universal Serial Bus (USB)
Also known asUSB · 通用串行总线 · Universal Serial Bus
Universal Serial Bus (USB) is a standard for connecting peripherals, defined by the USB Implementers Forum (USB-IF), carrying both data and power between computers and external devices. It unified how keyboards, storage, displays and more attach to a host, with USB Type-A and USB Type-C as familiar connector shapes.
Term details→160UP creator
Also known asUP主
UP creator is a common Chinese video-community label for people who publish content, particularly on Bilibili.
Term details→161User Experience (UX)
Also known asUX · 用户体验 · User Experience
User experience (UX) describes the entire process and overall feeling a person has when using a product, including usability, efficiency and emotional response; the term was popularized in the 1990s by cognitive scientist Don Norman. UI concerns how an interface looks and behaves, while UX concerns the whole journey and its outcomes.
Term details→162User Interface (UI)
Also known asUI · 界面 · 用户界面 · User Interface
A user interface (UI) is the surface where a human interacts with a machine, covering visible and operable elements such as buttons, menus, icons and text. A graphical user interface is one form of UI; command lines and voice interfaces are also UIs.
Term details→163Venture Capital (VC)
Also known asVC · 创投 · 风险投资 · Venture Capital
Venture capital (VC) is a form of financing in which professional investment firms put money into early-stage companies that have high growth potential — and a high risk of failure — in exchange for equity, often entering at stages such as seed or Series A.
Term details→164Vibe Coding
Also known as氛围编程
Vibe coding, coined by Andrej Karpathy in February 2025, describes a way of building software where you state what you want in natural language, let a large model generate the code, and iterate through conversation without reading every line. It speeds up prototyping but is debated for maintainability and security.
Term details→165Virtual Machine (VM)
Also known asVM · 虚拟机器 · 虚拟机 · Virtual Machine
A virtual machine (VM) is a software-based computer emulated by a virtualization layer (hypervisor) on a physical machine, running its own operating system and applications. Multiple VMs can share the resources of the same physical hardware while remaining isolated from one another.
Term details→166Virtual Private Network (VPN)
Also known asVPN · 虚拟专用网络
A VPN establishes a private logical connection over a public network, commonly using an encrypted tunnel to a remote network or gateway.
Term details→167Virtual Private Server (VPS)
Also known asVPS · VPS 主机 · 虚拟专用服务器 · Virtual Private Server
A VPS is a virtual machine carved out of a physical server and dedicated to a single user, with its own operating system, resource quota and administrator access, while the physical hardware is shared with other tenants.
Term details→168Virtual Reality (VR)
Also known asVR · 虚拟现实 · Virtual Reality
Virtual Reality (VR) is a technology that builds and presents an entirely synthetic environment, usually experienced through a head-mounted display, with the view changing in real time as the head and body move. Its hallmark is replacing the user's real sensory input.
Term details→169Visual Effects (VFX)
Also known asVFX · 视效 · 视觉特效 · Visual Effects
Visual effects (VFX) refers to creating or altering imagery in post-production through compositing, computer-generated imagery (CGI) and related techniques beyond what was captured on set, widely used in film and advertising. 3D computer graphics are one of its core tools.
Term details→170Web 2.0
Also known asWeb2.0
Web 2.0 is a concept popularized by O'Reilly Media around 2004 for the phase of the web marked by user-generated content, participatory communities and open collaboration between sites. The web shifted from static documents to platforms that users read from and write to.
Term details→171Web crawler
Also known as爬虫
A web crawler automatically discovers and retrieves web pages, often for search indexing or information collection.
Term details→172WebAssembly (Wasm)
Also known asWASM · WebAssembly
WebAssembly (Wasm) is a binary instruction format for a stack-based virtual machine that runs in sandboxed environments such as browsers at near-native speed. It lets code written in languages like C, C++ and Rust be compiled and used on the web and in server-side runtimes.
Term details→173Webhook
Also known as网络钩子
A webhook sends an HTTP request to a configured URL when an event occurs, allowing one system to notify another.
Term details→174Wi-Fi
Also known asWiFi · IEEE 802.11 · 无线局域网
Wi-Fi is a wireless local area network technology based on the IEEE 802.11 family of standards, with the Wi-Fi Alliance handling trademark and certification. It lets phones, laptops, and other devices connect through wireless routers and is among the most common ways to access networks at home and in public spaces.
Term details→175Workflow
Also known asWorkflow Automation
A process orchestrated through predefined steps; in the AI context, a workflow has developers prescribe the path and the LLM follow it, as opposed to an agent that decides its own steps.
Term details→176Worktree
Also known asgit worktree · 工作目录
A worktree is a working directory in Git with a branch checked out. With the git worktree command, one repository can have several branches checked out in different directories at the same time, each working independently while sharing the same object store.
Term details→177x86
Also known asx86-64 · AMD64 · IA-32 · x86 架构
x86 is a family of instruction set architectures originating from Intel's 8086 processor, characterized by variable-length instructions and long dominance of personal computers and servers. Later extensions, such as the 64-bit x86-64, grew the address space and instruction set while preserving software compatibility.
Term details→178Zero-day vulnerability
Also known as0-Day · Zero-Day · 零日漏洞
A zero-day vulnerability is a hardware, firmware, or software flaw unknown to defenders or lacking an available fix; exploiting it is a zero-day attack.
Term details→