AI Knowledge Base
From AI fundamentals and practical tools to reliable workflows.
Explore methods for learning, work and collaboration.
Explore five topics
Getting Started with AI
Start with a small task whose result you can check yourself.
Choose a task, explain it clearly and check the answerUsing AI Tools
Understand inputs, settings and limits, then work through practical tool problems.
Choose tools for your materials and troubleshoot their useAI at Work
Turn source material into presentations, notes and deliverables you can revise and hand over.
Produce useful, editable workAI Workflows & Teams
Organize processes, materials and responsibilities so work can be repeated and shared.
Make a one-off method repeatable and transferableAI Trends
Read technical and industry changes in context, separating events, evidence and interpretation.
Understand changes, evidence and their effect on workFeatured articles
Selected starting points, practical guides and methods.
Series & topic guides
All articles
Browse the library by topic.
- AI Workflows & TeamsHow to Use OpenAI dot for Work: Lessons from Real TasksLessons from using OpenAI dot for research, content and website work, with a reusable task brief, decision handoff and guidance for reviewing deliverables.
- Getting Started with AICodex Tutorial for Beginners: From Choosing an Interface to Finishing Your First TaskChoose a Codex interface, prepare a local folder, and turn fictional feedback into a checked handoff table. Includes practice files and translations of verified Chinese outputs.
- AI at WorkWhen you hand work to Codex, what can you deliver—and what must you still decide yourself?Choose manual work, chat, or Codex file-based work. Check sources, revisions, and handoffs while keeping unapproved decisions and commitments with the right people.
- Using AI Toolsvideo-use subtitle burn-in failed? Check the FFmpeg subtitles filterPreserve your generated SRT, identify the FFmpeg executable video-use actually uses, and check its subtitles filter before changing builds.
- Using AI ToolsChatGPT Attachments Missing from Project Sources After Moving a Chat?Find the correct original, check for duplicates, add it through Project Sources, and verify the contents. Know what to record if the process still fails.
- AI at WorkNarration and Scene Timing: A Manual Audio-First WorksheetPlan scene cuts from audio durations, distinguish separate chapter files from one continuous track, and review transitions. Includes a copyable blank worksheet and a fictional worked example.
- AI at WorkPPT Master: Edit Selected Slides While Preserving the DesignScope a selected-slide update with PPT Master’s documented Edit Native PPTX workflow. Confirm the plan and check untouched slides, shared resources, and export records.
- Using AI ToolsPresenton Custom Template 504: Separate Timeouts from Processing FailuresDiagnose a 504 during Presenton custom PPTX template import. Match browser and backend records, test a sanitized one-slide copy, and prepare a useful issue report.
- Using AI Toolshtml-video Exports Default Text? Compare the Preview and Render InputsTrace html-video preview and export mismatches in Studio or the CLI. Check saved HTML, template variables, versions, and output files, then collect a minimal report.
- Using AI ToolsChatGPT missed PDF table rows? Check pages, row counts, and fieldsBefore totaling a PDF table in ChatGPT, compare expected rows, source locations, and field values. Use a six-row example and a copyable prompt to check gaps.
- Using AI ToolsJianying Skill: Diagnose Missing Media in DraftsA source-based guide to missing media in Jianying Skill drafts: check paths, IDs, cloud music and macOS errors, then prepare a useful issue report.
- Using AI ToolsLibrary vs. SDK vs. Framework vs. API: Who Calls Whom?Compare libraries, SDKs, frameworks, and APIs by tracing who calls whom, how data moves, and where errors occur in a runnable Python exercise.
- Using AI ToolsFrom a Python Script to a Small Project You Can RerunStructure a small rerunnable Python project with an entry point, input, helper module, JSON persistence, and README, then verify the result from a fresh process.
- Using AI ToolsHow Does Python Process Data? From Variables and Containers to Conditions and FunctionsTrace Python data from values and types through containers, conditions, loops, functions, and return values so you can predict behavior, locate failures, and verify output.
- Using AI ToolsWhere Is Python Actually Running? Understanding Anaconda, Jupyter, Terminal, and EnvironmentsTrace Python from interface and shell to interpreter, kernel, environment, packages, and working directory so you can diagnose common setup failures with evidence.
- Getting Started with AILLMs Are Not Knowledge Bases: Why Next-Token Prediction Can Produce Confidently Wrong AnswersUnderstand why fluent LLM output is not factual verification, and when AI applications need external data, retrieval, tools, validation, permissions, or human review.
- Using AI ToolsIs a Longer Context Always Better? From Tokens and Context Windows to Context EngineeringLearn how tokens, context windows, state, memory, and context engineering fit together—and why better-selected context matters more than simply adding more tokens.
- Getting Started with AITransformer Without the Math: What Attention and Embeddings Actually DoUnderstand Attention, Embeddings, and KV Cache without deriving matrices. Connect context, semantic search, RAG, and inference cost for AI application engineering.
- Using AI ToolsPrompt Is Not Magic: Structured Output, Tool Calling, and NondeterminismLearn how prompt constraints, Structured Output, Tool Calling, temperature, top-p, and invariant-based testing turn an LLM into a more reliable AI application component.
- AI Workflows & TeamsThe Model Can Be Fine and the System Can Still Fail: What Does Production AI Reliability Actually Mean?Learn why model quality alone does not make an AI application reliable, and how retrieval, tools, validation, fallback, observability, latency, cost, and evaluation shape production outcomes.
- Getting Started with AIWhat Exactly Is an AI Agent? From Chatbots to Closed-Loop Decision SystemsLearn why an AI Agent is more than an LLM with tools, and how Goal, State, Observation, Decision, Action, and Feedback form a closed-loop system.
- AI Workflows & TeamsLLM, RAG, Workflow, or Agent? Choosing the Minimum Necessary ComplexityLearn when to use deterministic code, an LLM, Retrieval or RAG, a Workflow, an Agent, or Multi-Agent—and why the simplest architecture that satisfies the task is often the strongest choice.
- AI Workflows & TeamsAn Agent Is More Than an LLM: How State, Tools, and Memory Form a Working SystemLearn how State, Context, Memory, Retrieval, Tools, and the Agent Runtime form a working closed loop—and why system truth should live outside model-generated text.
- AI Workflows & TeamsAction Space, Permissions, HITL, and Stop Conditions: How to Give an Agent Capability Without Losing ControlLearn how Action Space, least privilege, runtime permissions, HITL, reversibility, budgets, and Stop Conditions keep an Agent capable without giving it unlimited agency.
- Using AI ToolsCC Switch Setup: Verify a Provider Change and Switch BackConfigure one provider in CC Switch, verify requests and billing records, and test restoring your previous Codex CLI configuration.
- AI at WorkAI Video Editing: Start with a Talking-Head Rough Cut and SubtitlesPlan a first AI-assisted edit with explicit cut decisions, subtitle timing checks and a reviewable rough cut. Understand external transcription requirements.
- Using AI ToolsCodex Permission Mode Unavailable: A Four-Layer Diagnostic ChecklistCheck your Codex environment, workspace scope, file and network boundaries, approval policy and managed profiles before changing settings.
- Using AI ToolsExport HTML Animation to MP4 with Playwright and FFmpegRender a six-second HTML animation at 24 fps, encode 144 frames with FFmpeg, and verify the dimensions, duration and animation timing.
- AI TrendsAI Technical Weekly | 2026-W37: Agent Permissions and Inference Routing Move into RuntimeGitHub agent permissions, SageMaker prefix-aware routing, and CUDA 13.4 show production AI control moving into runtime policy, cache locality, and GPU isolation.
- AI Workflows & TeamsHow to Build AI Long-Term Memory with a Memory RouterBuild a small Memory Router that controls scope, retrieval, context injection, and version status without loading an entire knowledge base into every AI conversation.
- AI at WorkHow to Create an Editable AI PowerPoint—and Prove It Is EditableA PPTX filename does not prove editability. Use this three-slide exercise to verify native text, chart data, separate images, save, and reopen behavior.
- Using AI ToolsCan Claude Watch Videos? A Transcript-and-Frame WorkflowLearn what a transcript can prove, when frames are needed, and how to request a timestamped video summary without pretending Claude received the full video.
- AI at WorkHow to Build Traceable AI Notes in ObsidianUse five portable Markdown files to preserve sources, separate judgment from evidence, limit what AI receives, and verify the answer chain.
- Using AI ToolsHow to Back Up and Restore Codex Work When Moving ComputersSeparate projects, configuration, conversation context, and credentials; then verify file integrity and a real work path in an isolated migration exercise.
- AI TrendsAI Technical Weekly | 2026-W36: GPT-6 Astra, Gemini 3.8 Flash, and Agentic VideoGPT-6 Astra, Gemini 3.8 Flash, and agentic video understanding bring distinct questions about runtime control, GA model evaluation, and demand-driven multimodal retrieval.
- AI TrendsAI Technical Weekly | 2026-W35: Inference Chips, Hardware Agents, and Multimodal ModelsOpenAI Jalapeño, Anthropic MHS, and Gemini Omni 1.1 Flash move inference efficiency, hardware-agent control, and multimodal workflows forward without proving production readiness.
- Getting Started with AIHow to Complete Your First Reviewable Task With CodexComplete your first Codex task with safe source material, a bounded request, and a review checklist that lets you verify every important result.
- Getting Started with AIShould You Add Context, Examples, or a Reference Format?Diagnose why an AI result missed the mark, then add only the context, examples, or reference format that solves the specific gap.
- Getting Started with AIHow to Fix an AI First Draft Without Starting OverKeep what works, describe the expected result, actual result, and gap, then run one focused revision you can review against the original task.
- AI at WorkHow to Check Whether AI Citations and Numbers Are RealVerify that each AI citation exists, locate the exact supporting passage, recalculate important numbers, and check the surrounding context before use.
- AI at WorkHow Should You Verify AI Facts, Inferences, Advice, and Decisions?Classify AI output before checking it: trace facts to evidence, test inferences, compare recommendations, and keep final decisions with a named person.
- Using AI ToolsHow ChatGPT Data Access Changes Across Personal, Managed Accounts, and AppsUnderstand who can access ChatGPT data, how managed accounts differ from personal accounts, what connected apps can receive, and what to check about retention and permissions.
- AI at WorkHow to De-Identify Data or Build Synthetic Practice MaterialIdentify direct and indirect identifiers, preserve only the task structure you need, and create synthetic examples that do not represent real people.
- AI Workflows & TeamsShould AI Work Live in a Project, Folder, or Conversation?Organize recurring AI work by duration, shared context, and saved outputs so projects, folders, and conversations each have a clear role.
- AI Workflows & TeamsWhich Instructions Belong in AGENTS.md?Save durable, scoped, testable project rules in AGENTS.md. Keep one-off deliverables, temporary preferences, and stale facts in the current task.
- AI Workflows & TeamsWhy Does AI Forget, and How Should You Manage Versions?Separate temporary context, retrievable memory, and authoritative files so AI work can survive long conversations without silently using stale versions.
- AI Workflows & TeamsWhat Is a Skill, and Which Work Is Worth Turning Into One?Use four checks—repetition, stable steps, defined inputs and outputs, and reviewable failure—to decide whether a workflow deserves a reusable Skill.
- Using AI ToolsWhat Problems Do Plugins, Connectors, and MCP Solve?Choose an extension by the missing capability: packaged behavior, access to external data, or a protocol that exposes tools and resources.
- Getting Started with AIWhat Is an AI Agent? Chat, Work, and Agentic Tasks ExplainedCompare chat, longer-form AI work, and agents by steps, tool use, external actions, and human control so you can choose the safer entry point.
- Getting Started with AIHow to Run a Safe First AI Agent ExperimentUse copied files and four observable stages—read, plan, act, review—to test an AI agent without external logins or irreversible actions.
- AI Workflows & TeamsHow to Design Agent Permissions, Approvals, Stops, and RecoveryLimit access first, require approval before consequential actions, define stop conditions, and verify a recovery path before the agent runs.
- AI Workflows & TeamsHow to Prove an AI Workflow Is Repeatable Before AutomationTest standard input, variations, missing data, errors, and same-input reruns. Record intervention and decide go or not yet before scheduling.
- AI Workflows & TeamsHow to Build a Team SOP From a Personal AI WorkflowDocument purpose, roles, inputs, steps, checks, exceptions, and version ownership, then test the SOP with someone who did not design it.
- AI Workflows & TeamsHow to Test Whether a Skill, Plugin, or Workflow Is ReliableWrite the expected behavior first, then test normal, boundary, failure, rerun, and permission cases with saved evidence and a clear decision.
- AI Workflows & TeamsHow Should Teams Assign Responsibility for AI Work?Use a simplified RACI matrix to assign data, process, access, approval, incident, and version responsibilities with explicit escalation paths.
- AI Workflows & TeamsHow to Design a Small, Measurable, Stoppable AI PilotChoose one low-risk use case, a small group, and a short period. Compare quality, time, revision, and failures, then decide go, revise, or stop.
- Getting Started with AIHow to Choose Your First AI Task: Start Small, Low-Risk, and ReviewableChoose a useful first AI task with four checks and a simple task selection card. Start with safe material, a reviewable draft, and no irreversible action.
- Getting Started with AIHow Does Generative AI Produce Answers—and Why Can It Be Confidently Wrong?Learn why generative AI can produce fluent but incorrect answers, then use a simple boundary map to separate drafts, facts, and decisions that need verification.
- Getting Started with AIChat vs ChatGPT Work vs Codex: What’s the Difference?Compare Chat, ChatGPT Work, and Codex by task: discussion, document and deliverable creation, or coding and technical changes. See which one to start with.
- Getting Started with AICodex for Beginners: Understand the Workspace, Conversation, File Changes, and Command ResultsStart using Codex with a read-only exercise that helps you distinguish the workspace, conversation, actual file changes, and command results.
- Using AI ToolsCodex Permissions: How Sandbox, Approvals, Folders, and Network Access WorkUnderstand what Codex can read, edit, run, and access online. Learn how sandbox, Ask for approval, folder boundaries, and network permissions work before granting more access.
- AI Workflows & TeamsA Repeatable AI Image Workflow: From Brief to Reviewed PromptTurn an AI image request into a repeatable workflow with a clear brief, human review gate, stop conditions, and reusable records—without promising unattended automation.
- Getting Started with AICan You Put Company Data into AI? Four Checks Before You Upload ItBefore you upload company data to AI, check the data, account and service, your authority to share it, and whether a smaller or synthetic alternative can do the job.
- Getting Started with AICan You Trust an AI Answer? Five Checks Before You Use ItA fluent AI answer is not automatically ready to use. Check facts, calculations, sources, judgments, and recommendations before relying on it at work.
- AI at WorkAI Meeting Notes: Turn Source Text into Decisions and Action Items You Can ReviewUse AI meeting notes to turn authorized source text into reviewable decisions, action items, owners, and unknowns—without inventing commitments, dates, or responsibility.
- Getting Started with AIHow to Write Better AI Prompts: Define the Work Before You Tune the PromptDefine the goal, usable context, constraints, and a done condition to turn a vague AI request into a first draft you can inspect.