Family History & Genealogy Terms

The vocabulary of the work you've been doing. Some of these are old terms with sharp meanings. Some are CM-specific framings of common concepts. All of them show up in chronicles, research summaries, and the conversations you're already having with other researchers.

Family History

The broader term for the work — researching ancestors, finding records, telling the story, sharing it with people who carry the same name. Family history includes genealogy but goes further: it cares about the story, not just the proof. Chronicle Makers serves family historians.

Genealogy

The narrower term — the documented chain of ancestry, the proof, the records, the citations. Genealogy is one part of family history. People who identify as genealogists tend to focus on records and accuracy; people who identify as family historians tend to focus on the broader story. The terms overlap, but the emphasis differs.

Chronicle

A finished, written piece about one ancestor or one chapter of a family's life — and honestly, more history than genealogy. A chronicle places the person inside the world they actually lived in. Every fact is contextualized: the war happening that year, the law that changed where they could live, the weather of the harvest that mattered, the rate of childhood death in their county. Most genealogists don't understand why contextualization matters. It's the thing that brings a person to life. A list of dates and places is a record. A chronicle is what makes a reader care about the person behind the dates.

Chronicles are the output Chronicle Makers exists to help you finish. Not a tree. Not a research log. A story your family will actually read because the person inside it feels real.

Proof Argument

A written case that ties evidence together to support a genealogical conclusion — usually identifying a person, establishing a relationship, or solving a brick wall. A proof argument names the question, presents the evidence, addresses conflicts, and states the conclusion. The Genealogical Proof Standard (GPS) is the most cited framework for what makes a proof argument hold up.

Research Summary

A document that captures everything you know about an ancestor — facts, dates, locations, sources, conflicts, gaps. It's not a draft of the chronicle. It's the spine the chronicle gets written from. Lab 2 in Chronicle Makers ends with a research summary you'll cite from forever.

Evidence Inventory

A structured list of every source you have for a research question — categorized by type, date, location, and what each source proves. The Evidence Inventory is what you build before drafting a proof argument. Without it, you're writing from memory; with it, you're writing from sources.

Scope Statement

A single sentence that names what one chronicle covers and what it deliberately leaves out. "This chronicle covers Horace Wilmer's 1950 bankruptcy and what it cost him, not his earlier dairy farming years." A scope statement is the first step of the STORI Method — and the single most useful sentence a family historian can write before opening a draft.

Threading Ancestor

The ancestor whose story carries through a longer project — a book, a series, a season of writing. Threading ancestors give a multi-chronicle project a spine the reader can follow. Most family historians pick a threading ancestor without naming the choice. Naming it makes the rest of the work easier.

Source Backbone

The smallest set of sources that can carry a chronicle — usually three to seven. Not every source you have. The ones you can't write the story without. Identifying the source backbone is what turns a research file into a writing project.

Brick Wall

A research question you've worked on long enough that you've stopped making progress. Brick walls are usually about missing records, conflicting evidence, or a name that doesn't survive in any of the obvious places. The genealogy world treats brick walls as failures; Chronicle Makers treats them as topics — sometimes the most interesting chronicle is about the gap itself.

Source, Information, Evidence — the three axes

Modern genealogical methodology, as defined by Elizabeth Shown Mills in Evidence Explained, separates three concepts that library science conflates. A single document is a source. The facts inside that document are information. What that information proves about a research question is evidence. The same record can be an Original source containing both Primary and Secondary information used as Direct or Indirect evidence — depending on what fact you're working with. That precision is why "primary source" and "secondary source" are not genealogically accurate terms. The categories below replace them.

Original Source

A record that captures information at its first written appearance — a vital record at the courthouse, an original will in the probate file, a census enumeration sheet, an original letter. The first time the assertion made it into print.

Derivative Source

A copy, abstract, transcription, or index of an original — a typed transcript of a will, an index to a county marriage record, an online database entry derived from a paper record. Derivatives can introduce errors during copying. Use them as finding aids; verify against the original when possible.

Authored Source

A work the author created by gathering, correlating, and interpreting multiple sources — a published family history, a compiled genealogy, a county history, a journal article. Authored sources reflect the author's analysis and conclusions. They are useful for clues and context, not as proof on their own.

Primary Information

Information given by someone with firsthand knowledge of what they reported — a parent reporting the date of their child's birth, a coroner stating the cause of death, a soldier filing a pension claim about his own service. Primary information is the strongest information class but does not guarantee accuracy.

Secondary Information

Information given by someone reporting what they learned from others — a grandchild reporting an ancestor's birthplace, a death certificate informant naming the deceased's parents fifty years after the fact, a census taker recording ages from a neighbor's memory. Most genealogical errors live in secondary information presented as fact.

Direct Evidence

Evidence that explicitly answers your research question. A death certificate that names parents directly answers "who were her parents." Direct evidence is rare, especially before 1900.

Indirect Evidence

Evidence that supports a conclusion without stating it directly. Land deeds, witness names, neighbor patterns, naming conventions. Most genealogy of the 18th and 19th centuries lives in indirect evidence. Building a case from indirect evidence is the work of a proof argument.

Negative Evidence

The absence of a record where one should exist — often as informative as the record's presence. A man missing from a tax list the year his neighbors paid suggests he died, moved, or wasn't subject to tax. Negative evidence is real evidence when you know what you should be finding.

Citation

The full reference that lets another researcher find the exact source you used. Author, title, publisher, date, repository, collection, page, accessed-on date. Citations are the difference between research that holds up and research that disappears. Chronicle Makers uses a streamlined Chicago/Evidence Explained style for citations — see the chronicle-citation-style skill.

Vital Records

The four legal records every researcher learns first: birth, marriage, death, and divorce. Vital records become more available as you move forward in time. Pennsylvania didn't require statewide vital registration until 1906; before that, it's church records, family bibles, and county filings.

Chronicle Makers Methodology

The named frameworks, tools, and concepts that make up the Chronicle Makers approach. You don't need to memorize the names. You'll see them across the labs, the Sprint, the book, and the chronicles — they're the shared vocabulary of finishers.

Chronicle Makers Method

The book, the methodology, and the system. The Method takes a family historian from "I have 20 years of research and nothing finished" to "I have a chronicle my family is reading." The Chronicle Makers Method book launches in May 2026; the methodology runs through the five labs, the Sprint, and the community.

The STORI Method

The five-step process for writing one chronicle from start to finish: Scope, Thread, Originate, Reflect, Inspire. STORI is the spine of Lab 3 and the 10-Day Chronicle Writing Sprint. Most family historians who follow it complete a chronicle within ten days. It is the most cited Chronicle Makers framework.

Scope — name what this chronicle covers in one sentence. Thread — find the single line that holds the story together. Originate — draft scaffolding from your research summary. Reflect — review the draft for voice, evidence, and shape. Inspire — share with family and the public.

The AIM Framework

The three-step method for working with AI on genealogy research: Assess, Interact, Measure. AIM replaces "trust but verify" — a phrase that sounds wise and contains no testable instruction.

• Assess — before you open the AI tool. What do you already know? What records do you have? What specific question are you asking? You set the terms of the conversation, not the AI.
• Interact — the conversation itself. You provide context, ask specific questions, and guide the AI's work. You are the researcher. The AI is the assistant.
• Measure — after the AI responds. Every factual claim, every date, every name, every citation gets measured against your original sources. The AI's output is a hypothesis. Your sources are the evidence.

AIM is taught in Lab 1 and referenced across every lab. The word is always "measure," never "verify." The posture is always evaluation, never trust.

The Five Labs

The Chronicle Makers curriculum, structured as five sequential labs:

• Lab 1: Technology Fundamentals — set up your AI tools and workspace.
• Lab 2: Organizing Essentials — organize research, build a source-backed research summary.
• Lab 3: Chronicle Writing — the 10-Day Sprint and the STORI Method.
• Lab 4: Brick Wall Research — AI-powered methods for the hardest research problems.
• Lab 5: Chronicle Book Method — turn chronicles into chapters and a published book.

The Chronicle Makers Sprint

The 10-Day Chronicle Writing Sprint — a structured writing cohort where members take one ancestor from research summary to finished chronicle in ten days. 95% of writers who follow the daily steps finish. The Sprint runs quarterly with live cohorts and a recorded course inside Lab 3.

Chronicle Compass

A free, public web tool that helps family historians figure out what to write next and where to start. Not an evaluator. Not a test. A recognition mirror — the Compass shows you what you've already built, where you're strong, and one specific next step. Launches at chroniclemakers.com/compass during NGS week, late May 2026. Find your direction. Start your chronicle.

Finisher Framework

The distinction between finishers and meanderers — the two ways a family historian can spend twenty years. Meanderers research forever and never close anything; finishers cross the line, hand the chronicle to someone, and start the next one. The Finisher Framework names the choice and gives the finisher posture a vocabulary. It is the operating posture of Chronicle Makers content.

Full Pantry

The framework for thinking about whether your research is ready to write from — not as a binary "enough or not enough," but as a pantry you can cook from. Full Pantry asks: do you have what you need to make this meal? Not every meal needs every ingredient. Some chronicles can be drafted with a thin pantry; some need it stocked. The framework lets you decide which kind of chronicle you're writing without paralysis.

Maker Showcase

The monthly community event inside Chronicle Makers where members share finished work — chronicles, books, gifts, projects. The Showcase replaces the old "Lab Wins" format. Hearing other members read finished work out loud is the single most reliable motivator in the community.

Founders

Members who joined Chronicle Makers during the founding window (closed April 30, 2026) at the $480 lifetime rate. 61 founders are grandfathered at lifetime access; new members join at $29/month or $199/year.

Chronicle Makers Genealogy Research Analyst Plugin

The Claude Code and Cowork plugin built by Chronicle Makers. Follows a GPS research methodology. Available for download. Built on FSL-1.1-MIT license. Distributed through GitHub, Skool, and chroniclemakers.com.

American Chronicles

The strand of Chronicle Makers content that ties family stories to the American 250th anniversary (1776–2026). Every American Chronicle places an ancestor inside the Revolutionary or early Republic era, using the 250th as a structural occasion for storytelling. The Chronicle Makers Method book launches as part of this strand.

AI Vocabulary

You don't need to memorize these. This is a reference — look things up when you encounter a term you don't recognize. The terms are organized by how close they are to your daily experience, not by technical depth.

What AI Is

Artificial Intelligence (AI)

A broad term for computer systems that can perform tasks that typically require human thinking — like understanding language, recognizing images, or making decisions.

Generative AI

AI that creates new content — text, images, audio, code — rather than just analyzing existing content. ChatGPT, Claude, and Gemini are all generative AI.

Large Language Model (LLM)

The specific type of AI behind tools like ChatGPT and Claude. Trained on enormous amounts of text to understand and generate human language.

Model

The trained AI program itself. When people say "the model," they mean the core AI — separate from the interface you use to talk to it.

Neural Network

The mathematical structure that lets AI learn patterns, loosely inspired by how human brains process information. All modern AI is built on these.

Natural Language Processing (NLP)

The field of computer science focused on helping machines understand human language — how we speak, write, and communicate.

Machine Learning

The broader field that includes AI. Any system that improves at a task by learning from data rather than being explicitly programmed.

How AI Learns

Training Data

The massive collection of text, books, articles, and websites an AI reads to learn language patterns. The quality and scope of this data shapes what the AI knows — and what it gets wrong.

Pre-training

The first, massive round of learning where the AI reads as much text as possible to learn general patterns of language and knowledge.

Post-training

What happens after pre-training: the AI is refined using human feedback, safety guidelines, and alignment techniques (like RLHF) to make it more helpful, honest, and safe.

Fine-tuning

Adjusting an already-trained AI to make it better at specific tasks or domains — like making a general AI more skilled at legal writing or medical questions.

Knowledge Cutoff

The date when an AI's training data ends. The AI has no reliable knowledge of events after this date unless it can search the web. Always ask when the cutoff is.

Bias

When AI unfairly favors or ignores certain perspectives because of patterns in its training data. AI reflects the biases present in the text it was trained on.

Dataset

The organized collection of text used to train the AI. Different datasets produce different capabilities and blind spots.

How You Talk to AI

Prompt

The question, instruction, or input you give to an AI tool. Better prompts produce better results — this is a skill, not a talent.

Context

The background information that helps AI understand what you're asking. More relevant context = more useful responses.

Context Window

The maximum amount of text an AI can hold in its working memory during a conversation. When you exceed it, the AI starts forgetting earlier parts of the conversation.

Token

A piece of text the AI processes — roughly 3/4 of a word. AI counts usage in tokens, not words. A 1,000-word document is about 1,300 tokens.

System Prompt / Custom Instructions

Behind-the-scenes instructions that shape how the AI behaves in a conversation. Some tools let you set these yourself (Claude Projects, ChatGPT Custom Instructions).

Conversation / Thread / Chat

A single exchange with AI that maintains context. Starting a new conversation means the AI forgets everything from the previous one (unless using Projects or similar features).

Zero-shot Prompting

Asking AI to do something without giving it any examples first. "Summarize this census record" — and trusting it to figure out the format.

Few-shot Prompting

Giving AI a few examples of what you want before asking it to do the task. "Here are two census summaries I wrote. Now summarize this third one the same way."

Iteration

Refining AI output through back-and-forth conversation. The first response is rarely the final one — you shape it through follow-up instructions.

Output / Response

Whatever the AI gives back after your prompt — text, analysis, a draft, a list, a table, code.

Temperature

A setting that controls how creative or predictable AI's responses are. Low temperature = more predictable. High temperature = more creative (and more likely to confabulate).

How AI Sees (Not Just Reads)

Vision / Image Recognition

AI's ability to look at and understand images — photographs, scanned documents, handwritten text, maps. Not all AI tools have this; the ones that do call it "vision."

Multimodal

AI that can handle multiple types of input — text, images, audio, video, documents — in the same conversation. Most major AI tools are now multimodal.

OCR (Optical Character Recognition)

Technology that converts images of text (scanned documents, photographs of pages) into actual text a computer can read and search. AI has made OCR dramatically better, especially for historical handwriting.

What Can Go Wrong

Hallucination

When AI generates information that sounds plausible but is factually wrong — invented dates, nonexistent sources, fabricated quotes. The most dangerous AI problem for genealogists.

Confabulation

A more accurate term for hallucination. The AI isn't lying — it's filling gaps in its knowledge with plausible-sounding fiction, the same way a human might misremember a story and fill in details that feel right.

AI Slop

Generic, overpolished AI-generated content that reads like no human wrote it. Characterized by phrases like "delve into," "it's important to note," "rich tapestry," and excessive enthusiasm about everything.

Prompt Injection

A security risk where malicious instructions are hidden inside content that AI reads — a web page, a document, an email. The AI follows the hidden instructions instead of yours. Especially dangerous with agentic AI tools that can take actions.

Data Privacy

What AI companies do with your conversations. Policies vary: some use your data to train future models, some don't. Read the privacy policy of any tool you paste family information into.

Tools, Tiers & Products

Chatbot

An AI tool you interact with through conversation — typing prompts and getting responses. ChatGPT, Claude, and Gemini are all chatbots (among other things).

Free Tier / Paid Tier

Most AI tools offer a free version with limits (fewer messages, older models) and a paid version ($20/month typical) with higher limits and better models. You can finish a chronicle on free.

API (Application Programming Interface)

A way for software to talk to AI directly, without a human typing in a chat window. Developers use APIs to build tools. You don't need to know how to use one — but when someone mentions it, that's what they mean.

Plugin

An add-on that gives an AI tool new capabilities — like connecting it to a specific database, file system, or workflow. Think of it as an accessory for the AI.

Skill

A pre-written set of instructions that teaches AI how to do a specific task well — like transcribing a historical document or formatting citations. Used in Claude Cowork and Claude Code.

Extension

Software that adds features to your web browser. Claude in Chrome is a browser extension. Extensions run inside your browser, not on a separate server.

Model Version

AI companies release updated models with different names (Claude Sonnet, Claude Opus, GPT-4, Gemini Pro). Newer versions are generally more capable but sometimes behave differently. The version matters.

Harness

The interface that runs the AI model. The model is the engine; the harness is the car you drive. ChatGPT.com is a harness for OpenAI's models. Claude.ai is a harness for Anthropic's models. Same model, different harness = different experience.

MCP (Model Context Protocol)

A standard way for AI tools to connect to other software — your files, databases, apps, calendars. Think of it like a universal adapter. When an AI tool supports MCP, it can plug into more things without custom coding.

Agentic AI

Agent / AI Agent

AI that can take actions — not just answer questions, but do things: send emails, fill forms, browse websites, create files, interact with other software on your behalf.

Agentic AI

The broad category of AI tools that can act autonomously — making decisions and executing tasks beyond just generating text in a chat window.

Agentic Browser

A web browser with a built-in AI agent that can navigate websites, click buttons, fill forms, and complete tasks for you. Examples: Claude in Chrome, ChatGPT Atlas, Perplexity Comet.

Autonomy Level

How much an AI can do on its own. The spectrum: Level 1 (Chat) — you ask, it answers. Level 2 (Tool Use) — it takes actions with your approval. Level 3 (Autonomous) — it makes decisions independently. Level 4 (Agent Swarms) — multiple agents working together. We're at Level 3 now.

Tool Use

When AI can interact with external tools — searching the web, reading files, running code, calling other software — instead of just generating text.

Permission Model

How agentic AI asks for your approval before taking actions. Good tools propose an action, wait for your yes, then execute. Always read what it's proposing before approving.

Guardrails

Safety limits built into AI tools to prevent harmful or unintended actions. These are features, not bugs — they're the reason Claude sometimes says "I'd rather not do that."

Working With Files & Knowledge

Knowledge File

A document you give to AI that contains information about your research, your ancestor, or your project. It gives AI the context it needs to help you well.

Project (Claude Projects, Custom GPTs)

A persistent workspace where AI remembers your context across multiple conversations. Upload files once, reference them in every chat — instead of re-explaining your research every time.

Upload / Attachment

Sending a file (PDF, image, document) to an AI tool so it can read and analyze the contents. Not all tools support all file types.

Markdown

A simple way to format text using plain characters — # for headings, ** for bold, - for lists. AI tools read and write it natively. Easy to learn, useful everywhere.

Plain Text vs. Rich Text

Plain text has no formatting (just letters and numbers). Rich text has fonts, colors, and layout (like a Word document). AI works best with plain text and markdown.

PDF

AI can read most PDFs, but scanned PDFs (images of pages) require OCR first. If you paste text from a PDF and it looks garbled, the PDF is probably a scanned image, not real text.

AI Safety & Design

Constitutional AI

How Claude is built. Instead of just training on human feedback, Claude follows a written set of principles — a "constitution" — that guides its behavior. It's why Claude sometimes pushes back, asks clarifying questions, or declines a request. It's not broken; it's designed to have values.

Alignment

The field of making AI systems do what humans actually want — safely, reliably, and without unintended consequences. This is why AI companies spend so much effort on testing and safety.

Copyright

When AI helps you write, the legal ownership of the output is still evolving. Generally: your original ideas and structure are yours. AI-generated text alone has uncertain copyright status. The safest approach is to always revise AI output substantially and make it your own.

Terms You'll Hear But Don't Need to Use

You don't need these to finish your chronicle. They're here so you recognize them when someone drops them in a Facebook group or a tech-heavy blog post.

Transformer

The underlying architecture (blueprint) behind most modern AI. When someone says "transformer-based model," they mean it uses the same fundamental design as ChatGPT and Claude. You don't need to know how it works.

Embedding

A way of turning words into numbers so the computer can do math with them. This is how AI understands that "dog" and "puppy" are related concepts.

Attention

The mechanism that helps AI figure out which words in a sentence are most important for understanding meaning. It's why AI can handle long, complex text.

Parameter

The internal settings a model adjusts during training. More parameters generally means more capable — GPT-4 has hundreds of billions. The number gets thrown around in marketing.

Inference

When AI uses what it already learned to answer your question. Every time you prompt ChatGPT or Claude, you're running inference. You'll see it on pricing pages.

Supervised Learning / Unsupervised Learning

Two ways AI can learn. Supervised: learning from labeled examples ("this is a cat, this is a dog"). Unsupervised: finding patterns without labels. Pre-training is mostly unsupervised; post-training is mostly supervised.

RAG (Retrieval-Augmented Generation)

AI that looks up information from a specific source before answering, instead of relying only on its training. When someone says "RAG-powered," they mean the AI is referencing real documents, not just its memory.

Latent Space

The internal representation where AI stores its understanding of concepts. Very abstract, very overhyped in tech discussions. When someone says "in the latent space," they're being unnecessarily technical.

Vector / Vector Database

The mathematical format AI uses to store and compare concepts internally. A vector database stores these for fast lookup. You'll see this in product marketing — it means the tool can search through documents quickly using AI.

CLI (Command Line Interface)

A text-based way to interact with your computer by typing commands instead of clicking icons. Claude Code runs in a CLI. Most people never need this.

GUI (Graphical User Interface)

The visual version — windows, buttons, menus, icons. Everything you normally interact with on your computer is a GUI. When someone says "it has a GUI," they mean you don't need to type commands.

IDE (Integrated Development Environment)

A workspace where developers write and test code. Claude Code runs inside IDEs. You don't need one for chronicle writing — this is a developer tool.