- Published on
Lesson 2 of "How to Use AI for Family History"
In Lesson 1, we built your prompt foundation: Role + Context and Task + Guardrail. You learned to treat AI like a hyper-fast research intern, adopted one privacy habit (no details about living relatives), and picked up the 3-minute check for anything an AI tells you.
Now we put that formula to work on a question that trips up many researchers: getting to know the place where your ancestor lived. Geography is the most common roadblock in family history. Records are filed by jurisdiction (the government office that had authority over a place), and jurisdictions change, so a search in the wrong town or county can look like a brick wall when it's really just the wrong door.
By the end of this lesson, you'll have built three practical tools: a Town Profile (a summary of the place: which county or other government covered it, how that changed, and what happened there), a Records Roadmap (a table of who kept the records and where they are held today), and a Map-Finder (a prompt that tells you which historical maps to look for and what to search for). Each tool comes in two depths: a short version for the Beginner track and a detailed version for the Intermediate track. You'll also learn two upgrades to your prompt formula, and where AI should stop and real maps should take over.
Bring from Lesson 1: your finished "try it first" prompt for your track and the AI's reply, saved in a document and tagged "lead, unverified."
Why the Place Matters
Ancestors didn't live in a vacuum. Before you search for a single record, read the layers of the place: when it was settled and incorporated, what larger jurisdiction it belonged to, how its borders moved, which neighbors and churches mattered, and what happened there.
Here is why that matters. If an ancestor's farm never moved but the county line was redrawn right over it, searching today's county seat leads straight to a dead end. The same thing happens everywhere. In colonial New England, a town may have belonged to a different county than it does now. In Ireland, records are often organized by townland, civil parish, or barony (small local land divisions). In Central Europe, a village may have changed countries more than once, and its records may now sit in a different country's archives. Almost anywhere, records can be split across local, regional, and national archives.
Doing this by hand means digging through county histories, old reference books, and archive websites, and it can eat up entire weekends. AI can give you a head start, with one important condition, which comes next.
The Right Job for AI: Study Buddy, Not Encyclopedia
An encyclopedia is a trusted reference book, and an AI chat tool isn't one. When you ask an AI about a place, you are not looking up an answer. You are asking an assistant to draft a first-pass briefing: a study guide that tells you what to look for and where.
Then you check that study guide against the real references:
- town and county histories
- state, provincial, or national archive catalogs
- the FamilySearch Research Wiki
- a guide to your region's records, such as The Genealogist's Handbook for New England Research
Think of it this way: the AI drafts the briefing, and you and the reference books decide what's true.
Level Up Your Prompt Formula: Add Format, Add Qualifiers
Lesson 1's formula still works. Lesson 2 adds one new part and strengthens another:
ROLE + CONTEXT + TASK + FORMAT (new) + [ GUARDRAIL ] (w/qualifiers)
= A TOWN PROFILE YOU CAN CHECK
New Part: Format
Tell the AI what shape you want back. Without a format, you get a chatty essay you have to dig through. With one, you get something you can scan, verify line by line, and save to your notes.
For locality work, a table is ideal. For example: "Present your answer as a table with these columns: Years | Jurisdiction | Records created | Where held today | How I can verify it."
Guardrail, level two: add qualifiers
In Lesson 1, the guardrail said "don't invent anything." Now we add three qualifiers. Qualifiers are short instructions that limit how sure the AI is allowed to sound, and they make it show its work:
- Confidence labels. Ask it to mark each claim High, Medium, or Low confidence and say what would confirm it.
- Named sources. Ask it to name the act, history, or record it is relying on, and to say so if it can't name one.
- Ranges, not guesses. Ask for a range when a date is uncertain, instead of a single "exact" year.
Two cautions. First, avoid asking a chat tool to "cross-reference" documents unless it can search the web, because it may pretend to. Second, a named source is a lead, not a fact. AI can produce plausible-sounding titles that don't exist, so look every source up yourself.
Start Your Lesson Notes
From now on, every lesson ends with a short list of what to save. Those saved pieces are your lesson notes, and the next lesson starts from them. If you've never kept research notes, this is an easy way to begin.
- Create one document. Use a word processor, a notes app, or anywhere you can paste text. Name it "AI Family History: Lesson Notes."
- Add a heading for each lesson. Start now: paste your Lesson 1 prompt and the AI's reply under "Lesson 1."
- Date every entry and name the tool. Write the date and the AI tool you used. Replies differ between tools and change over time, so this tells you which one you're reading.
- Tag every entry. "Lead, unverified" means the AI said it and you haven't confirmed it. "Checked" means you confirmed it against a source, and you write that source beside it. Nothing goes in untagged.
- Paste, don't rely on the chat. Copy the reply into your document. Chat histories can vanish or be hard to search.
These notes hold the lesson's worked example from your chosen track. If you try the Takeaway Prompts on your own ancestors, keep that work in its own document so the lesson notes stay clean.
Step One: Meet the Town (Town Profile)
Before any records search, learn the place. Each track follows its own family. Beginners build a short Town Profile for the Harlow family in Worcester, Massachusetts, around 1880. Intermediate researchers build a detailed one for the Calder family in Lancaster, Massachusetts, from 1650 to 1730. The questions are the same for any place: which jurisdiction held the records then, and where are they now? Think of building your Town Profile as reading the layers: it surveys the ground before anyone picks up a shovel. Each prompt below is labeled by formula part so you can see the formula at work.
BEGINNER TRACK: a short Town Profile
Goal: build a short Town Profile that says which county Worcester belonged to in 1880 and today, and whether it was a town or a city.
Why it matters: beginners often spend hours searching today's county or a state-level database for records that were kept by a city clerk, a county office, or a church instead.
Prompt:
ROLE
Act as a patient, beginner-friendly genealogy guide.
CONTEXT + TASK
Example: I am researching the Harlow family, who lived in Worcester, Massachusetts, around 1880. Tell me what county Worcester belonged to then and what county it belongs to today, whether it was a town or a city in 1880, and give me a short history of the place up to that year.
FORMAT
Use a short list with plain-language headings.
GUARDRAIL
Mark anything you're unsure about, tell me how I can confirm each fact myself, and do not invent facts.
Answer key: a good reply should say that Worcester was first settled in 1673 as Quinsigamond Plantation, was abandoned in 1675 during King Philip's War, and was permanently resettled in 1713. It was incorporated as a town in 1722, became the shire town (county seat) of the new Worcester County in 1731, and was incorporated as a city in 1848. In 1880, Worcester was a city in Worcester County, the same county as today. So the county didn't change, but the kind of government did, and that matters for which office kept the records. If the AI names a different county or a different year for the city charter, check a Worcester history or the Newberry Library's Atlas of Historical County Boundaries. Sources can also disagree on smaller details, such as the date of the second settlement attempt, and catching a disagreement is part of the skill.
INTERMEDIATE TRACK: a detailed Town Profile with a jurisdictional timeline
Goal: a detailed Town Profile covering settlement and incorporation, parent jurisdictions in each period, boundary changes and towns carved out, neighbors, church or parish organization, and local events that could have affected the records.
Why it matters: historical records follow the jurisdiction that existed at the time, and local history explains the gaps. Lancaster is a good example. Most of the town was destroyed during King Philip's War in 1676, and resettlement began in 1679, so a missing decade in the records may have a historical cause. Wars, partitions, and settlement waves leave gaps in every region, so ask the AI which events shaped your place.
Prompt:
ROLE
Act as a professional genealogist specializing in 17th- and 18th-century New England jurisdictional history and archival methodology.
CONTEXT + TASK
Example: I am researching the Calder family in Lancaster, Massachusetts, between 1650 and 1730. Build a detailed Town Profile covering when the area was settled and incorporated, the parent jurisdictions it belonged to in each period, any boundary changes or towns carved from it, neighboring towns, church or parish organization, and major local events that could have affected record keeping.
FORMAT
Present a table with these columns: Years | Jurisdiction (town, parish, county, state, province, or country) | What changed | Confidence | How I can verify it.
GUARDRAIL
Label each claim High, Medium, or Low confidence and name the act, history, or record you are relying on. If you cannot name one, say so. If a date is uncertain, give a range. Do not cite compiled genealogies as proven fact, and do not guess.
Answer key: a strong reply should include the 1643 settlement, the 1653 incorporation, the move into the new Worcester County in 1731 from Middlesex County, the 1676 destruction and 1679 resettlement, and later towns set off from Lancaster (Harvard in 1732, Bolton in 1738, Leominster in 1740, among others). Anything it adds beyond that is a lead to verify, not a finding.
Step Two: Follow the Paper Trail (Records Roadmap)
Once you've checked your Town Profile, don't start from scratch. Copy the AI's reply from Step One, with your corrections, and paste it into a second prompt that asks where the records live. Think of it as building on what you've checked.
Why this step comes second: repositories are organized by jurisdiction. You can't ask "where are the records?" until you know which jurisdictions existed. In colonial Massachusetts, for example, the town clerk, the county (deeds, probate, courts), the colony or province, and the local church each created different records, and each may be held somewhere different today. Worcester in 1880 followed the same pattern. The city clerk kept births, marriages, and deaths, the county's registry of deeds and probate court kept land and wills, the state kept its own vital-record copies after 1841, and churches kept their own registers. Every place has its own mix, so the question travels: who kept which records, and where are they now?
One rule applies to both tracks: paste in only a Town Profile you have checked. If you paste in an unchecked one, you are building on guesses.
BEGINNER TRACK: a short Records Roadmap
Goal: use your short Town Profile from Step One to ask which offices or organizations kept the basic records (birth, marriage, and death) and where they are held now. The reply is a short Records Roadmap: a small table.
Prompt:
ROLE
Act as a patient, beginner-friendly genealogy guide.
CONTEXT + TASK
Example: Here is a short Town Profile of Worcester, Massachusetts, that I have checked: [paste your checked Town Profile from Step One]. Based on it, tell me which offices or organizations kept birth, marriage, and death records for the Harlow family in Worcester around 1880, and where those records are held today. Give me 2 to 3 specific repositories as names I can search for.
FORMAT
Present a table with these columns: Record type | Who kept it then | Where it is held today | How I can verify it.
GUARDRAIL
Do not invent archives, and say "I'm not sure" wherever you aren't.
Compare: Hold this reply beside the one you saved in Lesson 1. It's the same family and the same question. What did the checked Town Profile and the table format change?
INTERMEDIATE TRACK: a detailed Records Roadmap
Goal: ask for a detailed Records Roadmap across every jurisdiction in your Town Profile, with record groups, date ranges, finding aids (the catalogs and guides to a collection) to check first, and a confidence label on each row.
Prompt:
ROLE
Act as a professional genealogist specializing in 17th- and 18th-century New England jurisdictional history and archival methodology.
CONTEXT + TASK
Example: Using the checked Town Profile, create a detailed Records Roadmap for Lancaster, Massachusetts, between 1650 and 1730: [paste your checked Town Profile from Step One]. For each jurisdiction in the Town Profile, list the record groups it created (town or parish records, land, probate, tax, court, church, or your region's equivalents), the years covered, where those records are held today, and which finding aids or catalogs to check first.
FORMAT
Present a table with these columns: Jurisdiction | Record group | Years | Where held today | How I can verify it | Confidence. Label each row High, Medium, or Low confidence.
GUARDRAIL
Note any known record loss, but flag any loss claim you are not certain of. Do not cite compiled genealogies as proven fact, and do not guess.
Compare: Hold this reply beside the one you saved in Lesson 1. It's the same family and the same question. What did the checked Town Profile and the table format change?
Step Three: Hunt for Historical Maps (Map-Finder)
Geography is the one area where AI helps least. A chat tool can describe borders in words, but it cannot reliably draw or place historical boundaries, and a confidently described line can simply be wrong. So in this step you use AI only to decide which maps to look for and what to search for, then use real map collections to see the maps themselves. The real maps you locate become your map sources, and you will keep them in your notes beside your checked Town Profile.
Where to find real historical maps:
- Atlas of Historical County Boundaries (Newberry Library): documents the creation and changes of U.S. counties from the colonial period on. It covers U.S. counties only, so for town-level changes (in New England, for example) you will still need town histories. For other countries, look for national or regional place-name guides and boundary atlases, and ask the AI what exists for your area. Then confirm each one is real.
- Library of Congress, Geography and Map Division: digitized maps, atlases, and town and county maps.
- David Rumsey Map Collection: a large online collection of historical maps.
- Town and county histories, deeds, and plats: these often include maps or surveys of the land itself.
BEGINNER TRACK: Map-Finder
Prompt:
ROLE
Act as a patient, beginner-friendly genealogy guide.
CONTEXT + TASK
Example: I am researching the Harlow family in Worcester, Massachusetts, around 1880. Tell me what kinds of historical maps would show this place at that time, and what search terms I should use to find them.
FORMAT
Use a short list with two plain-language headings: Kinds of maps and Search terms.
GUARDRAIL
Do not describe or draw the borders. Only name a specific map or collection if you are sure it exists, and tell me how I can confirm it. Otherwise, tell me what to search for.
Locate: Take one search term to a collection in the list above and open a real map of Worcester. Write down its title, its date, and where you found it. Save it to your Map sources list in your notes.
Compare: Lay the map beside your checked Town Profile. Do the borders and dates agree? If not, you haven't failed. You've found a research question.
INTERMEDIATE TRACK: Map-Finder
Prompt:
ROLE
Act as a professional genealogist and map librarian.
CONTEXT + TASK
Example: I am researching the Calder family of Lancaster, Massachusetts, between 1650 and 1730. Using my checked Town Profile to decide which jurisdictions to look under, list the types of maps and surveys (town maps, county atlases, plats, military maps) that would show this place in that period, what each would tell me about boundaries, roads, mills, meetinghouses, and neighbors, and which online collections or archives to search.
FORMAT
Present a table with these columns: Map type | What it can show | Years it covers | Where to search | Search terms | Confidence | How I can verify it.
GUARDRAIL
Label each row High, Medium, or Low confidence. Do not draw or describe borders from memory. Flag any collection you cannot confirm exists, and do not name a specific map unless you can say where it is held.
Locate: Open at least two maps from different decades or makers. Record each map's title, maker, date, and holding collection.
Compare: Lay the map beside your checked Town Profile. Do the borders and dates agree? If not, you haven't failed. You've found a research question.
After You Press Enter: The Reality Check
Remember the 3-minute check from Lesson 1: confirm one lead exists, look for sources, and label what the AI told you as unverified. For locality work, the same three steps get a geography twist. Don't save a Town Profile or Records Roadmap in your notes without a tag, and don't trust a map suggestion until you've opened the map. Run each through these checks first.
BEGINNER TRACK: check your Town Profile, Records Roadmap, and map sources
Checks:
- Confirm the repository exists. Find it on its own website or in the FamilySearch Catalog. The FamilySearch Research Wiki page for your town or county is a good place to see which record types exist and where they are held.
- Check the holdings. Does the archive's catalog actually cover the years you need?
- Go old school. Archives move collections, and town clerk offices keep limited hours. Check the official website or pick up the phone before you plan a trip.
- Check the county and dates. Confirm each formation or boundary date in a town history or the Newberry Library's Atlas of Historical County Boundaries.
- Check the map. Confirm the map or collection exists on the holder's own website or in a catalog, and note the map's date. If the AI named a specific map you can't find, treat it as invented.
INTERMEDIATE TRACK: check your Town Profile, Records Roadmap, and map sources
Checks:
- Confirm the repository exists. Find it on its own website or in the FamilySearch Catalog. The FamilySearch Research Wiki page for your town or county is a good place to see which record types exist and where they are held.
- Check the holdings. Does the archive's catalog actually cover the years you need?
- Go old school. Archives move collections, and town clerk offices keep limited hours. Check the official website or pick up the phone before you plan a trip.
- Apply the two-source rule to every date. Confirm each formation or boundary date in a second, independent source, such as the Newberry Library's Atlas of Historical County Boundaries, a town history, or a research guide for your region. If two sources disagree, record both and note the conflict.
- Spot-check one record-loss claim. These are a favorite spot for AI to invent, so confirm at least one with the repository.
- Check each map twice. Confirm it exists and note its maker and date, then apply the two-source rule to any boundary you take from it.
- Keep tidy notes. Tag unchecked items 'lead, unverified.' Move only checked facts into your Town Profile, each with its source, and list each map you opened under it as a map source.
Save to Your Notes: Lesson 2
You won't run the Map-Finder again in later lessons, but what it leads you to belongs in your notes, so keep a short list called "Map sources" under your checked Town Profile.
Beginner: your checked short Worcester Town Profile, your Records Roadmap, and the map you opened, listed under "Map sources." In Lesson 3 you can add one sentence from the profile to your prompt.
Intermediate: your checked detailed Lancaster Town Profile; your Records Roadmap with its confidence labels; your "Map sources" list, with title, maker, date, holding collection, date you looked, and one line on whether each map agrees with your Town Profile. Tag any map you haven't opened "lead, unverified." In Lesson 3 you'll use these maps to test the landmarks in a will.
Your Takeaway Prompts
Copy, paste, and fill in the bracketed blanks with your own ancestor details. Run Prompt A first and read the reply against your references. Fix anything that is wrong, then copy the corrected reply and paste it into the bracket in Prompt B. That reply is your checked Town Profile. Prompt C tells you which historical maps to look for. Once you have located real maps, list them under your checked Town Profile as your map sources. (This third step adds the most time, so Prompt C can wait until after Prompts A and B are checked.)
BEGINNER TRACK: building a Town Profile, Records Roadmap, and Map-Finder
Prompt A: Town Profile
Act as a patient, beginner-friendly genealogy guide. My ancestor lived in [Town/City, State or Country] around [Year]. Tell me what county that place belonged to then and what county it belongs to today, what kind of local government it had then (town, city, or something else), and give me a short history of the place up to that year. Use a short list with plain-language headings. Mark anything you're unsure about, tell me how I can confirm each fact myself, and do not invent facts.
Prompt B: Records Roadmap
Act as a patient, beginner-friendly genealogy guide. Here is a short Town Profile of my ancestor's town that I have checked: [paste your checked Town Profile]. Based on it, tell me which offices or organizations kept birth, marriage, and death records for this place around [Year], and where those records are held today. Give me 2 to 3 specific repositories as names I can search for. Present a table with these columns: Record type | Who kept it then | Where it is held today | How I can verify it. Do not invent archives, and say "I'm not sure" wherever you aren't.
Prompt C: Map-Finder
Act as a patient, beginner-friendly genealogy guide. I am researching [Town/City, State or Country] around [Year]. Tell me what kinds of historical maps would show this place at that time, and what search terms I should use to find them. Use a short list with two plain-language headings: Kinds of maps and Search terms. Do not describe or draw the borders. Only name a specific map or collection if you are sure it exists, and tell me how I can confirm it. Otherwise, tell me what to search for.
INTERMEDIATE TRACK: building a Town Profile, Records Roadmap, and Map-Finder
Prompt A: Town Profile
Act as a professional genealogist specializing in [Region, e.g., 17th- and 18th-century New England] jurisdictional history and archival methodology. I am researching a family in [Town/City, State or Country] between [Start Year] and [End Year]. Build a detailed Town Profile covering when the area was settled and incorporated, the parent jurisdictions it belonged to in each period, any boundary changes or towns carved from it, neighboring towns, church or parish organization, and major local events that could have affected record keeping. Present a table with these columns: Years | Jurisdiction (town, parish, county, state, province, or country) | What changed | Confidence | How I can verify it. Label each claim High, Medium, or Low confidence and name the act, history, or record you are relying on. If you cannot name one, say so. If a date is uncertain, give a range. Do not cite compiled genealogies as proven fact, and do not guess.
Prompt B: Records Roadmap
Act as a professional genealogist specializing in [Region] jurisdictional history and archival methodology. Using the checked Town Profile below, create a detailed Records Roadmap for [Start Year] to [End Year]: [paste your checked Town Profile]. For each jurisdiction in the Town Profile, list the record groups it created (town or parish records, land, probate, tax, court, church, or your region's equivalents), the years covered, where those records are held today, and which finding aids or catalogs to check first. Present a table with these columns: Jurisdiction | Record group | Years | Where held today | How I can verify it | Confidence. Label each row High, Medium, or Low confidence. Note any known record loss, but flag any loss claim you are not certain of. Do not cite compiled genealogies as proven fact, and do not guess.
Prompt C: Map-Finder
Act as a professional genealogist and map librarian specializing in [Region]. I am researching [Town/City, State or Country] between [Start Year] and [End Year]. Here is my checked Town Profile: [paste your checked Town Profile]. Using it to decide which jurisdictions to look under, list the types of maps and surveys (town maps, county atlases, plats, military maps, or your region's equivalents) that would show this place in that period, what each would tell me about boundaries, roads, mills, churches, and neighbors, and which online collections or archives to search. Present a table with these columns: Map type | What it can show | Years it covers | Where to search | Search terms | Confidence | How I can verify it. Label each row High, Medium, or Low confidence. Do not draw or describe borders from memory. Flag any collection you cannot confirm exists, and do not name a specific map unless you can say where it is held.
Coming up in Lesson 3:
You now know how to build a Town Profile and a Records Roadmap, and what historical maps to look for. Keep your checked Town Profile in your notes with your map sources listed beneath it. You'll paste the profile into prompts in the lessons ahead, and you'll lean on the maps when a boundary or land description comes up.
But once you find the right archive and open a digitized collection, whether it's a dense 19th-century census or a handwritten 17th-century town book, how do you sift all that messy text without getting overwhelmed?
In Lesson 3, we'll explore how to use AI to sort the names, dates, and relationships in a typed record into a clean, checkable table, without letting it interpret the evidence for you. The practice records are supplied.
Bring to Lesson 3:
- Beginner: your checked short Worcester Town Profile, your Records Roadmap, and the map you opened, listed under "Map sources." In Lesson 3 you can add one sentence from the profile to your prompt.
- Intermediate: your checked detailed Lancaster Town Profile, your Records Roadmap, and your "Map sources" list, saved in your notes.
What is the most confusing boundary shift or locality hurdle you've run into in your family history research? Drop a comment below!
