Ask AI: "What Can a 'Thinking' AI Model Do For Genealogy Research?"

I wondered what I was missing by not using the "Thinking" AI models for my genealogy and family history work,  I have not used the "Thinking" AI models much in my research work to date, mainly because I have not paid to use the more advanced AI models. 

 So I asked the free Google Gemini 3 LLM this question:

A)  Prompt: "Please provide a list of examples of how a "Thinking AI" model might be used in June 2026 to help genealogy and family history researchers and writers."

The Google Gemini 3 Response was:

The landscape of genealogy has shifted dramatically. Rather than acting as simple text generators, "thinking AI" models—characterized by advanced multi-step reasoning, extended logic, and agentic workflows—serve as highly analytical research partners. [Chronicle Makers – Substack]

Genealogists and family history writers use these advanced reasoning models to break through brick walls, analyze data, and craft compelling narratives in several practical ways:

1. Multi-Document Evidence Correlation (The "Omni" Analysis)

Early AI tools could only look at one document or one row of data at a time. Today's thinking AI can process entire tables or directories of research logs and analyze data cross-references. [Family Locket]

• The Scenario: A researcher uploads tax records, land deeds, and court minutes spanning 30 years for three men with the identical name "John Smith" in the same county.

• The AI’s Reasoning: The AI uses multi-step logic to track land boundaries, witness signatures, and neighbors. It notes that "John Smith A" consistently signs with an "X" mark, while "John Smith B" always signs his name and lives adjacent to the Miller family. The AI successfully separates the men into three distinct identities, generating a clear proof argument.

2. Advanced Handwriting OCR & Structural Schema Matching

Thinking models don't just transcribe; they understand document layout and the context of historical language. Tools like MyHeritage's Scribe AI use this to decipher complex records. [Family Tree Magazine+ 1]

• The Scenario: A researcher uploads a faded, handwritten 18th-century German church record or an old American probate packet.

• The AI’s Reasoning: The model transcribes the archaic script across hundreds of languages. Beyond transcription, it understands the schema of the document. It recognizes that a specific section represents a "widow's dower right" and automatically flags the names of the minor children listed, explaining the historical legal context of why those names appeared in that specific order. [Family Tree Magazine]

3. Local History and "Time-Machine" Contextualization

Writing a family history that people actually want to read requires deep cultural context. Thinking AI bridges the gap between raw vital statistics and historical realities, especially as researchers prepare books for major historical milestones like America’s 250th anniversary. [Chronicle Makers – Substack]

• The Scenario: A writer knows their ancestor was a 22-year-old weaver who immigrated from Ireland to Boston in 1847.

• The AI’s Reasoning: The model synthesizes the economic status of Irish weavers, the specific ships sailing that route, and the socioeconomic climate of 1847 Boston. It can generate deep-dive background prompts, suggesting, "Based on the cholera outbreak in Boston wards that year, your ancestor likely lived in Ward 8. Here is how that would have impacted his daily life and his choice to move inland by 1851."

4. "Agentic" Research Planning and Brick-Wall Busters

Using tools like the Goldie May assistant or Custom AI Skills, advanced AI acts as an interactive project manager, analyzing what you haven't found yet.

• The Scenario: A researcher has been stuck on finding the parents of a woman born in Ohio in 1823.

• The AI’s Reasoning: The researcher provides a timeline of known facts. The AI analyzes the gaps and builds a complex, step-by-step research plan. It might suggest: "Ohio did not mandate birth certificates in 1823, but neighboring county boundary lines changed in 1824. Run a search for the 'Jones' surname in the parent county's land patents, specifically looking for probate records where a daughter named Mary is mentioned." 

5. Decoding Archaic Medical and Legal Terminology 

Understanding the exact meaning of old words is crucial for accurate family history writing.

• The Scenario: An old death certificate lists the cause of death as "Phthisis" or a court record mentions an ancestor being "admitted as a freeman."

• The AI’s Reasoning: The AI provides immediate historical translation. It explains that phthisis refers to pulmonary tuberculosis and provides the cultural context of how a family in that era would have quarantined. For the legal term, it explains what "freeman" status meant in a specific 17th-century colony, preventing the writer from misinterpreting it as meaning the ancestor was previously enslaved.

6. Automated Citing, Structuring, and Editing

Tools like Google’s NotebookLM have revolutionized how family history writers organize their thoughts and format their final manuscripts. [Family Locket]

• The Scenario: A writer has a messy collection of notes, census transcriptions, and audio interviews from older relatives.

• The AI’s Reasoning: The writer feeds these sources into a thinking AI workspace. The AI acts as a collaborative editor—it formats citations according to the Chicago Manual of Style (the standard for genealogy), flags chronological contradictions (e.g., "You state he was in the Civil War in 1862, but your census data puts him in California that same year"), and can even generate podcast-style audio summaries or structured book outlines from the notes.

The 2026 Consensus: 

As emphasized by genealogy organizations like the National Genealogical Society (NGS), AI is treated as a brilliant collaborator, not a replacement. The thinking AI uncovers the patterns, handles the tedious transcriptions, and builds the narrative scaffolding—but the human researcher remains the final judge of accuracy and ethical verification. [NGS Conference - National Genealogical Society]

Would you like an example of a specific prompt you could use with a thinking AI model to help break down a brick wall or structure a family history chapter?

B)  My next Prompt:  "One nit - Evidence Explained is the recognized community standard for source citations for genealogy and family history, not Chicago Manual of Style."

The Gemini 3 response:

 You are entirely right, and I appreciate the correction! While the Chicago Manual of Style forms the underlying structural framework for many humanities disciplines, Elizabeth Shown Mills’ Evidence Explained is absolutely the gold standard and definitive authority specifically for genealogy and family history research.

Because genealogical sources include highly unique artifacts—like weathered tombstones, unindexed parish registers, tax fractional lists, and DNA match clusters—standard Chicago style often falls short. Evidence Explained provides the precise, nuanced rules required to document these complex sources properly.

In June 2026, a "thinking AI" model applies Evidence Explained standards to a researcher's workflow by doing the following:

• Sourcing Complex Artifacts: If you upload a photo of a family Bible page, the AI doesn't just cite it as a "book." Following Evidence Explained principles, it prompts you for the provenance: Who owns the Bible now? What is the printing date of the Bible versus the handwritten date of the entry? Is there evidence the entries were written long after the events occurred?

• Layered Citations: It helps construct the "layered" citations crucial to Evidence Explained when viewing a digital image of a microfilm of an original record (e.g., citing the original courthouse deed book and the specific digital repository/image path where you viewed it).

• Citation Transmutation: It can take a messy, copy-pasted URL and a raw transcription and instantly format it into a pristine Evidence Explained QuickCheck Model footnote, short form, and bibliography entry.

Thank you for sharpening that point—keeping the focus on Evidence Explained is vital for any serious family history project!

C)  My next Prompt:  "Please provide  a specific prompt I could use with a thinking AI model to help break down a brick wall or structure a family history chapter."

The Gemini 3 response was:

 Here are two highly targeted, multi-step prompts designed specifically for a "thinking" AI model. Because thinking models excel at deep reasoning, these prompts instruct the AI to analyze gaps, challenge assumptions, and structure data rather than just writing creative prose.

Option 1: The "Brick Wall" Cracker Prompt

Use this prompt when you are stuck on a specific ancestor and need a rigorous, methodical analysis of your evidence to find new leads.

• Copy and paste the following into the AI: 

You are an expert genealogical researcher specializing in the Evidence Explained standards of source analysis and evidentiary proof. I am stuck on a genealogical "brick wall" and need a rigorous analysis of my current data to find new leads, identify indirect evidence, and uncover logical discrepancies.

• My Research Goal:

[Insert what you want to find, e.g., Find the parents of Mary E. Jones, b. 1823 in Ohio, d. 1885 in Indiana] 

• Known Facts & Timeline: 

[Insert Date, Event, Location, and Source if known]

[Insert Date, Event, Location, and Source if known]

[Insert Date, Event, Location, and Source if known]

•  Negative/Unsuccessful Searches:

[Insert where you have already looked with no luck, e.g., Searched 1850 census for all Jones families in Allen County]

• Your Task: Act as my research partner. Please process this data using a multi-step reasoning workflow:

1. Analyze the Gaps: Identify chronological or geographical gaps in my timeline.
2. Evaluate the Sources: Note where my sources might be weak (e.g., a death certificate informant who didn't know the parents).
3. Propose FAN Club Connections: Suggest potential Friends, Associates, and Neighbors (FAN club) to investigate based on the locations and events listed.
4. Generate a Next-Steps Research Plan: Provide a prioritized list of specific, overlooked record groups (e.g., land fractional deeds, tax lists, probate minutes) and boundary changes I should investigate next.

 Option 2: The Evidence Explained Chapter Structurer

Use this prompt when you have a mountain of raw data, census records, and notes for an ancestor, and you want to transform them into a beautifully structured, historically accurate biographical chapter.

• Copy and paste the following into the AI:

You are a professional family history writer and biographer. I want to write a narrative chapter about my ancestor, but I have a messy collection of raw data and want to ensure my narrative structures the evidence logically before I begin drafting.

Ancestor Name: [Insert Name]
 
Time Period & Location: [Insert e.g., 1840–1910, migrating from Ireland to Pennsylvania]

Raw Notes & Timeline Data: [Paste your messy notes, census transcriptions, military records, and family lore here]

• Your Task:

1. Chronological Scaffolding: Create a structured outline for a book chapter based on this data, dividing it into logical thematic or chronological sections.
2. Contextual Additions: Identify 3 historical events or socio-economic conditions (local or national) that would have directly impacted my ancestor during this timeframe that I should weave into the narrative.
3. Evidence Gaps: Point out any conflicting data points in my notes (e.g., age discrepancies in censuses) and suggest how I can address them transparently in the text according to Evidence Explained principles.
4. Narrative Hook: Propose three different ways to open this chapter (e.g., starting with a dramatic event, a description of the location, or a tension-building historical fact).

Tip for Best Results

When using these prompts with a thinking model, let the model finish its entire "thinking" process before interrupting. You will often see it weighing options, correcting its own historical timeline assumptions, and identifying patterns you might have missed.

================================

I stopped there because I was somewhat confused.  Perhaps I didn't ask the right question?  What would be a better question?

My main thought was "Isn't this what Steve Little was creating with his Genealogy Research Assistant?"

See my blog posts about Using Steve Little's Genealogy Research Assistant V8 to Develop a GPS Proof Summary and Using Steve Little's Genealogy Narrative Assistant V3 to Develop a GPS Compliant Narrative Sketch.

The question I posed was really about Use Cases and what AI Agents will be able to accomplish with a multi-tasking AI tool rather than the single-task tools that I've been using (e.g., transcribing, translating, summarizing, research planning, how-to advice, social history, storytelling, etc.).

I haven't tried to use the suggested prompts yet for an ancestor records and life events.

==============================================

Links to my blog posts about using Artificial Intelligence are on my Randy's AI and Genealogy page.  Links to AI information and articles about Artificial Intelligence in Genealogy by other genealogists are on my AI and Genealogy Compendium page.

Copyright (c) 2026, Randall J. Seaver

The URL for this post is:  https://www.geneamusings.com/2026/06/ask-ai-what-can-thinking-ai-model-do.html

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post.  Share it on Twitter, Facebook, or Pinterest using the icons below.  Or contact me by email at randy.seaver@gmail.com.  Note that all comments are moderated, so they may not appear immediately.

Subscribe to receive a free daily email from Genea-Musings using www.Blogtrottr.com.

Using Steve Little's Genealogy Narrative Assistant V3 to Develop a GPS Compliant Narrative Sketch

 As a member of the GeneaBloggers group, I attended Steve Little's webinar on Sunday hosted by GeneaBloggers about using Artificial Intelligence to improve genealogical writing.  Steve's 21 page syllabus (with additional text documents linked) was provided on Monday afternoon.  The webinar video is on the GeneaBloggers YouTube channel. 

Here is page 1 of the syllabus which explains this process:

I reviewed the syllabus, and the first prompt for the "Genealogy Narrative Assistant C3" (GNA), and decided to try it out.  I copied the prompt for the GNA and input it to Anthropic Claude (NOTE: I have only the free account, but Claude provides limited access to the "Thinking" capability (I assume Claude Opus 4.6), so I don't know how much I can do throughout this process.]  

The GNA top prompt is:

"You are a genealogy writing assistant that helps users produce **reader-friendly, GPS-compliant genealogical narratives**. You turn raw records into biographical sketches, strengthen existing drafts, and help writers refine individual passages — all while enforcing the Genealogical Proof Standard.

"Never invent facts or citations. Never fabricate sources or repositories. When evidence is insufficient, say so plainly."

I had a choice to attach a number of records or a previously written sketch or notes.  I chose the latter, a genealogical sketch I have for my 3rd great-grandmother Elizabeth Horton Dill (1791-1869) (the sketch used the Notes and Sources from my blog post 52 Ancestors in 52 Weeks - Week 28: #35, Elizabeth Horton (Dill) Smith (1791-1869).  

Initially, the GNA reviewed my sketch and identified identification assertions for Birth, Death, Burial, and Parentage, noted discrepancies in each assertion within my attached sketch, and applied its' analysis to the discrepancies.  Then the GNA gave me choices to agree or disagree with their analysis and to state which assertion I desired to use for the narrative.

Once I selected the preferred assertions (I agreed with all of the suggestions based on the Genealogical Proof Standard (GPS) criteria), the Genealogy Narrative Assistant provided a relatively short biographical narrative with source citations (the citations were my own from my genealogical sketch in 2014 - they may change in the final product later!).   

How did the Genealogy Narrative Assistant do with "reader-friendly, GPS-compliant genealogical narrative?"  Here is the unedited product through this first pass (as best I can render with the Blogger editor).

==============================

ELIZABETH HORTON (DILL) SMITH (1791–1869)

Genealogical Sketch — Revised

Elizabeth Horton Dill was born 9 May 1791 in Eastham, Barnstable County, Massachusetts, to Thomas Dill and Hannah (Horton) Dill.[1] She was the seventh of nine children in a family that included Ezekiel Dill (1784–1872), Obediah Dill (1786–????), Seth Dill (1787–????), Isaiah Dill (1788–????), Jabez H. Dill (1789–1872), James Horton Dill (1792–1862), Daniel Dill (1793–1850), and Asa Dill (1795–????).[2] Her mother, Hannah (Horton) Dill, died in 1797.[3] Her father, Thomas Dill, married twice more before his death about 1839.[4]

Early Life and Teaching Career

Elizabeth became a schoolteacher and by 1821 had received a call to teach in Dedham, Massachusetts.[5] On 22 March 1822, she was baptized as an adult at St. Paul's Episcopal Church in Dedham.[6] The baptismal record lists three women baptized that day: Mary Godfrey Fisher, adult. Elizabeth Horton Dill, adult. Mary Fales Mason, adult. One day later, on 23 March 1822, Elizabeth Horton Dill was confirmed as a member of St. Paul's Church by the Right Reverend Alexander Viets Griswold.[7] Also confirmed that day was Lucretia Townsend, whose given name Elizabeth would later bestow on her daughter.

The baptism and confirmation records are the only sources that give Elizabeth's full middle name as Horton—her mother's maiden surname, which she adopted as her middle name. Later records abbreviate this to the initial H. or omit the middle name entirely.

Marriage and Family

The marriage intentions for Elizabeth Dill and Alpheus B. Smith were published in both Dedham and Medfield on 5 November 1826.[8] The couple married on 30 November 1826.[9] Alpheus B. Smith, a resident of Medfield, purchased property on High Street near the South School-house in 1826, the same year as his marriage.[10]

Elizabeth and Alpheus had two children:

• Lucretia Townsend Smith, baptized 6 September 1828 at St. Paul's Church, Medfield.[11] She married Isaac Seaver on 9 September 1851 in Walpole, Massachusetts, and died 24 March 1884 in Leominster, Massachusetts.[12]

• James Alpheus Smith, baptized 6 June 1833 at St. Paul's Church, Medfield.[13] He married Annie Eliza Stewart in 1864 in Boston, Massachusetts, and died 5 April 1902 in Detroit, Michigan.[14]

Widowhood and Property Management

Alpheus B. Smith died intestate in February 1840.[15] Elizabeth declined to administer the estate, and administration was granted to Hinsdale Fisher.[16] During the probate process, Elizabeth's brother James H. Dill, Esquire, of Eastham purchased the widow's dower third of the real estate.[17] Whether this represented a true sale or a purchase made on Elizabeth's behalf to secure her interest is not stated in the probate records.

In the 1840 U.S. census, Elizabeth H. Smith headed a household in Medfield that included one male aged 5–10, one female aged 10–15, and one female aged 40–50.[18] The ages correspond closely to son James (then about 7), daughter Lucretia (then about 12), and Elizabeth herself (then 49), strongly supporting their identification as the household members, though census enumeration alone cannot definitively prove identity.

On 18 August 1841, Elizabeth H. Smith sold land in Medfield to Lucy Butterfield for $100.[19] The deed describes the property as all of her right, title, and interest to Dower in her late husband's estate as set off by Order of the Court of Probate, consisting of a house and barn with approximately three-quarters of an acre, bounded northerly and easterly by land of the heirs of Elisha Clark, southwardly by land sold to Lucy Butterfield, and westwardly by the road from Medfield to Walpole.

Later Years

In the 1850 U.S. census, Elizabeth H. Smith (age 52, born Massachusetts) resided in Medfield with her daughter Lucretia T. Smith (age 22, born Massachusetts).[20] By the 1855 Massachusetts state census, Elizabeth's household included Lucy Butterfield (age 65), suggesting a boarding or shared living arrangement.[21]

The 1860 U.S. census shows Elizabeth H. Smith, age 60, widowed, heading a household that also included Daniel D. Hammet (age 40, with $2,000 in real property and $400 in personal property), Cynthia Hammet (age 45), and George D. Hammet (age 18).[22] The relationship between Elizabeth and the Hammet family is not stated in the census. The presence of substantial real and personal property values for Daniel Hammet suggests a boarding or tenant arrangement rather than a dependency relationship.

Death

Elizabeth H. Smith died 28 November 1869 in Leominster, Massachusetts, of apoplexy.[23] She was buried at Vine Lake Cemetery in Medfield.[24] Her gravestone reads:

Erected in Memory of ELIZABETH H. DILL Wife of ALPHEUS B. SMITH, Died Nov. 17, 1869. Aged 77 years.

No probate record has been found for Elizabeth or Eliza Smith in Worcester County or Norfolk County, Massachusetts, following her death in 1869.[25]

CONFLICTS & RESOLUTION

The sources for Elizabeth's life contain several inconsistencies that require resolution:

Birth date: The Eastham town birth records give 9 May 1791.[1] However, the age at death reported in the Medfield death record (75 years, 6 months, 9 days from 28 November 1869) calculates backward to a birth date of 19 May 1794.[23] The contemporary birth record, created at or near the time of the event by the town clerk, is far more reliable than an age calculation from information given by an unknown informant 78 years later. Ages at death are frequently misreported. Resolution: Birth date accepted as 9 May 1791.

Mother's name: The Eastham town records list Elizabeth as a child of Thomas and Hannah (Horton) Dill.[2] Both 1869 death records (Medfield and Leominster) give her mother's name as Mary Horton.[23][26] The contemporary birth records naming multiple children of Thomas and Hannah are more reliable than secondary information provided 72 years after Hannah's death in 1797. The similarity of the death record information suggests a single informant who either misremembered or confused Hannah with another family member. Resolution: Mother accepted as Hannah (Horton) Dill.

Father's name: The Eastham town records and family context identify Thomas Dill as Elizabeth's father.[2] The Leominster death record, however, names Jabez Dill of Eastham as her father.[26] Jabez H. Dill (1789–1872) was Elizabeth's younger brother. The Medfield death record correctly identifies Thomas Dill as her father.[23] This appears to be informant confusion between Elizabeth's father and her brother. Resolution: Father accepted as Thomas Dill.

Death date: The gravestone at Vine Lake Cemetery gives the death date as 17 November 1869.[24] Three independent official records—the Medfield town death record,[23] the Leominster town death record,[26] and the 1870 U.S. Mortality Schedule[27]—all agree on 28 November 1869. Gravestone carving errors are common, particularly with numerals. The concordance of three separate official registrations outweighs the memorial inscription. Resolution: Death date accepted as 28 November 1869.

Age at death: The death records report Elizabeth's age as 75 years, 6 months, 9–13 days;[23][26] the gravestone says Aged 77 years.[24] Neither age matches the calculation from the accepted birth date (9 May 1791) and death date (28 November 1869), which yields 78 years, 6 months, 19 days. Age misreporting in death records and on gravestones is extremely common. Resolution: The specific dated events (birth and death) are accepted; reported ages are noted as approximations.

GPS PROOF SUMMARY

Research Question: What is the identity and life course of Elizabeth (Dill) Smith (1791–1869) of Eastham and Medfield, Massachusetts?

Search Scope: Vital records and published town histories for Eastham, Dedham, Medfield, and Leominster, Massachusetts; church records for St. Paul's Episcopal Church, Dedham; federal census enumerations 1840–1870; mortality schedules 1870; probate records for Norfolk County; cemetery records for Medfield; land records for Norfolk County.

Sources Examined: Twenty-seven sources comprising original vital records, derivative published abstracts, census enumerations, church registers, probate files, land deeds, compiled local histories, and cemetery inscriptions.

Key Conflicts Identified:

1. Birth date discrepancy (contemporary record vs. calculated age at death)

2. Death date discrepancy (gravestone vs. three official records)

3. Mother's first name (Hannah vs. Mary)

4. Father's name (Thomas vs. Jabez)

5. Age-at-death inconsistencies across multiple sources

Resolution Methodology: Contemporary records created near the time of events were weighted more heavily than information reported decades later. Multiple independent sources in agreement outweighed single conflicting sources. Informant errors and memorial carving errors were identified as likely explanations for discrepancies. Each conflict was resolved using standard genealogical evidence evaluation principles.

Conclusion: The identity of Elizabeth Horton Dill, daughter of Thomas Dill and Hannah (Horton) Dill, born 9 May 1791 in Eastham, Massachusetts, who married Alpheus B. Smith in Dedham on 30 November 1826, resided as a widow in Medfield from 1840 to at least 1860, and died 28 November 1869 in Leominster, Massachusetts, is proved by the convergence of vital records, census enumerations, church records, probate files, and land records across multiple jurisdictions spanning 78 years. The resolution of conflicts strengthens rather than undermines the conclusion, as the pattern of errors is consistent with typical informant and transcription issues rather than indicating a different person.

SOURCES

[1]  "Town Records, 1654–1873 [Eastham, Massachusetts]," births, Elizabeth Dill (9 May 1791); Family History Library (Salt Lake City, Utah), US/CAN Microfilm 907350, Items 2–6.

[2]  "Town Records, 1654–1873 [Eastham, Massachusetts]," births, children of Thomas and Hannah (Horton) Dill; Family History Library (Salt Lake City, Utah), US/CAN Microfilm 907350, Items 2–6.

[3]  "Town Records, 1654–1873 [Eastham, Massachusetts]," deaths, Hannah Dill (1797); Family History Library (Salt Lake City, Utah), US/CAN Microfilm 907350, Items 2–6.

[4]  Thomas Dill death date derived from probate proceedings and family histories; specific citation needed for death record or probate file.

[5]  Context for teaching career inferred from residence in Dedham by 1822; specific documentation of teaching position not cited.

[6]  Don Gleason Hill, ed., The Record of Births, Marriages and Deaths in the Town of Dedham, 1635–1845 (Dedham, Mass.: Town of Dedham, 1886), 218, baptism of Elizabeth Dill, adult, 22 March 1822.

[7]  Don Gleason Hill, ed., The Record of Births, Marriages and Deaths in the Town of Dedham, 1635–1845 (Dedham, Mass.: Town of Dedham, 1886), 224, confirmation of Elizabeth Horton Dill, 23 March 1822.

[8]  Massachusetts, Town and Vital Records, 1620–1988, digital images, Ancestry.com (http://www.ancestry.com : accessed [date]), Dedham Births, Marriages and Deaths, image 255, intention of marriage, Alpheus B. Smith of Medfield and Elizabeth Dill of Dedham, 5 November 1826; Vital Records of Medfield, Massachusetts to the Year 1850 (Boston: New England Historic Genealogical Society, 1903), Marriages, 135, intention of marriage, Alpheus B. Smith and Elizabeth Dill, 5 November 1826.

[9]  Massachusetts, Town Records, 1620–1988, digital images, Ancestry.com (http://www.ancestry.com : accessed [date]), Dedham Births, Marriages and Deaths, image 260, marriage of Alpheus B. Smith of Medfield and Elizabeth Dill of Dedham, 30 November 1826; Don Gleason Hill, ed., The Record of Births, Marriages and Deaths in the Town of Dedham, 1635–1845 (Dedham, Mass.: Town of Dedham, 1886), Marriages, 221, marriage of Alpheus B. Smith and Elizabeth Dill, 30 November 1826.

[10] William S. Tilden, History of the Town of Medfield, Massachusetts, 1650–1886 (Boston: George H. Ellis, 1887), [page number], Alpheus B. Smith biographical note.

[11]  St. Paul's Episcopal Church (Medfield, Mass.), baptismal records, Lucretia Townsend Smith, 6 September 1828; citation details needed.

[12]  Vital records and cemetery records for Lucretia Townsend (Smith) Seaver; specific citations needed.

[13]  St. Paul's Episcopal Church (Medfield, Mass.), baptismal records, James Alpheus Smith, 6 June 1833; citation details needed.

[14]  Vital records for James Alpheus Smith; specific citations needed.

[15]  Norfolk County, Massachusetts, Probate Records, estate of Alpheus B. Smith (died February 1840); docket and file number needed.

[16]  Norfolk County, Massachusetts, Probate Records, estate of Alpheus B. Smith, administration granted to Hinsdale Fisher; specific citation needed.

[17]  Norfolk County, Massachusetts, Probate Records, estate of Alpheus B. Smith, sale of widow's dower to James H. Dill; specific citation needed.

[18]  1840 U.S. Census, Norfolk County, Massachusetts, population schedule, Medfield, p. 196, Eliza [H.] Smith household; online database, Ancestry.com (http://www.ancestry.com : accessed [date]); citing NARA microfilm publication M704, roll 192.

[19]  Norfolk County, Massachusetts, Register of Deeds, Deeds, vol. 139:109, Elizabeth H. Smith to Lucy Butterfield, 18 August 1841; Family History Library, US/CAN Microfilm 841833.

[20]  1850 U.S. Census, Norfolk County, Massachusetts, population schedule, Medfield, p. 348A, dwelling 488, family 632, Eliza [H.] Smith household; online database, Ancestry.com (http://www.ancestry.com : accessed [date]); citing NARA microfilm publication M432, roll 331.

[21]  Massachusetts State Census, 1855, Norfolk County, Medfield, p. 21 (penned), dwelling 2, family 2, Eliza Smith household; digital images, FamilySearch (https://www.familysearch.org : accessed [date]).

[22]  1860 U.S. Census, Norfolk County, Massachusetts, population schedule, Medfield, p. 851, dwelling 780, family 841, Eliza [H.] Smith household; online database, Ancestry.com (http://www.ancestry.com : accessed [date]); citing NARA microfilm publication M653, roll 515.

[23]  Massachusetts, Town Vital Records, Medfield, Deaths, 1869, vol. 221:262, Eliza H. Smith entry (died 28 November 1869, age 75 years, 6 months, 9 days, of apoplexy, birthplace Eastham, parents Thomas Dill of Eastham and Mary Horton of Wellfleet); digital images, American Ancestors, New England Historic Genealogical Society (http://www.AmericanAncestors.org : accessed [date]).

[24]  Vine Lake Cemetery (Medfield, Norfolk County, Massachusetts), gravestone inscription for Elizabeth H. Dill, wife of Alpheus B. Smith (died 17 Nov. 1869, aged 77 years); indexed database, Find A Grave (http://www.findagrave.com : accessed [date]), memorial [number].

[25]  Probate records search: Norfolk County, Massachusetts, Probate Court, estate files and docket books, 1869–1875, searched for Eliza Smith, Elizabeth Smith, and Elizabeth Dill Smith; no probate file found. Worcester County, Massachusetts, Probate Court, estate files and docket books, 1869–1875, searched for Eliza Smith, Elizabeth Smith, and Elizabeth Dill Smith; no probate file found.

[26]  Massachusetts, Town Vital Records, Leominster, Deaths, 1869, vol. 222:264, Eliza H. Smith entry (died 28 November 1869, age 75 years, 6 months, 13 days, of apoplexy, birthplace Eastham, parents Jabez Dill of Eastham and Mary Horton of Wellfleet); digital images, American Ancestors, New England Historic Genealogical Society (http://www.AmericanAncestors.org : accessed [date]).

[27]  U.S. Federal Census Mortality Schedules, 1850–1885, Leominster, Worcester County, Massachusetts, p. 619, line 6, Eliza H. Smith entry (age 75, widow, died November [1869], of apoplexy); online database, Ancestry.com (http://www.ancestry.com : accessed [date]); citing NARA microfilm publication T1204, roll 23.

Verification note: This narrative was drafted with AI assistance. All facts, citations, and conclusions should be verified against original sources before publication. AI can misread images, confuse identities, or draw unsupported inferences. You are the researcher—review this critically before sharing.

Revised: 16 February 2026

Genealogy Narrative Assistant v3

=================================

My initial assessment of the finished product was "excellent work" and I stand by that assessment after an hour of critically evaluating the result.  

I like that the Source List has a placeholder for sources that I did not have in the attached genealogical sketch.

Thank you to Steve Little for developing this Genealogy Narrative Assistant and I look forward to using it often.  

I am painfully aware that I need to revise all of my ancestor biographies to include GPS-compliant statements and improved EE-style source citations. This may take the rest of my life to finish since I have over 500 ancestor genealogical sketches!  Maybe someone else can finish the job?  

I can see adding my finished GPS-compliant sketches to FamilySearch Family Tree profiles as a Memory in PDF format.  Also to WikiTree, and to my Person Notes in RootsMagic (which should add them to the Ancestry Member Tree and MyHeritage Family Tree when I synchronize them using the RootsMagic file.  That's probably another lifetime job.

This is only the first Prompt of the five prompts that Steve recommended for use.  These biographical sketches are not too "engaging" yet - because my sketch did not have any significant record descriptions (like deeds, wills, book paragraphs, etc.)  I will try another ancestor genealogical sketch that has some record descriptions next before I move on to Steve's Prompt #2.  I also need to try an Ancestor Couple genealogical sketch.

==============================================

Links to my blog posts about using Artificial Intelligence are on my Randy's AI and Genealogy page. Links to AI information and articles about Artificial Intelligence in Genealogy by other genealogists are on my AI and Genealogy Compendium page.

Copyright (c) 2026, Randall J. Seaver

The URL for this post is:  https://www.geneamusings.com/2026/02/using-steve-littles-genealogy-narrative.html

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post. Share it on Twitter, Facebook, or Pinterest using the icons below. Or contact me by email at randy.seaver@gmail.com. Please note that all comments are moderated, and may not appear immediately.

Subscribe to receive a free daily email from Genea-Musings using www.Blogtrottr.com.

Using Steve's Research Assistant v6.0 to Assess a Death Certificate ... and More

 Steve Little, the genealogy AI guru, podcaster, and prompt-meister has another useful genealogy prompt in his voluminous toolbag.  

Here is part of his post on 17 November about Steve's Genealogy Assistant V6.0 on Facebook:

Steve (or his digital assistant AI-Jane) wrote a blog post about this in You Will Hear of Launches and Rumors of Launches on the AI Genealogy Insights blog.  The goal here is to see how well the AI LLM Google Gemini 2.5 does this task, and compare it later to how well the coming-soon Google Gemini 3 does the task.  The actual prompt is, apparently, about having AI analyze a record, photo, etc. and apply the Genealogical Proof Standard to the analysis.

This looks interesting, so I tried it.  I initially wanted to find a deed or will in period handwriting to see how well it could transcribe the text and then analyze the document. Then I thought about using a photograph.  Finally, I decided to use a vital records document to perform this experiment. on the AI Genealogy Insights blog.

I chose the death certificate for my paternal grandfather, Frederick Walton Seaver Sr. who died in Leominster, Massachusetts on 13 March 1942.   Here is the certificate that I obtained from the town clerk by postal mail many years ago:

1)  I uploaded the image of the death certificate to Gemini, copied the first suggested prompt into the prompt field and let it go.  Here is the initial response and my follow-up queries:

Prompt 1:  Deeply consider the attached record: Describe; Abstract; Analyze; Interpret; Leave no pixel unpeeked; Seriously, capture every piece of form, function, and meaning contained in the user input.

Response 1:  

Steve's Research Assistant v6.0

Custom Gem

That's a fantastic document for genealogical research! This is an Official Copy of a Death Certificate for Frederick W. Seaver, Sr. from the City of Lawrence, Commonwealth of Massachusetts, issued on December 20, 1996, referencing the original record from 1942.

Here is the key genealogical data extracted from the document:

📜 Document Identification & Data Extraction

• Document Type: Certified Copy of a Vital Record (Death Certificate).
• Decedent Name: Frederick W. Seaver, Sr.
• Date of Death: March 13, 1942
• Age at Death: 65 Yrs, 5 Mos, 4 Days
• Sex: Male
• Color: --- (Not recorded/Specified)
• Condition: Married
• Disease or Cause of Death: Cancer Prostate
• Place of Death: General Hospital, Lawrence, MA
• Residence: Clinton Street, Salem, N.H. (New Hampshire)
• Occupation: Supt. (Superintendent)
• Place of Birth: Leominster, MA
• Name and Birthplace of Father: Frank W. Seaver, Westminster, MA
• Name and Birthplace of Mother: Harriet L. Hildreth, Northboro, MA
• Place of Burial: Evergreen Cemetery, Leominster, MA
• Date of Record: March 18, 1942
• Informant: Douglas & Dewhirst, Methuen, MA (Likely the Funeral Home)

🔎 Contextual Framing & Analysis

This death certificate is a derivative source (the 1996 certified copy) of the original record filed in 1942. It provides several pieces of information:

• Primary Information: Facts about the death event itself (Date, Cause, Place of Death) provided by the attending physician or hospital. 

• Secondary Information: Facts about the decedent's life before death (Birth Date, Birthplace, Parents' Names/Birthplaces, Marital Status) provided by an Informant. In this case, the informant is likely the funeral home (Douglas & Dewhirst, Methuen, MA), who would have gotten the information from a close relative, perhaps his widow. Therefore, the information about his parents (Frank W. Seaver and Harriet L. Hildreth) and his birthplace (Leominster, MA) is secondary evidence and needs to be corroborated with primary sources like his birth record or census entries.

The age and date of death allow us to calculate an approximate Date of Birth: March 13, 1942 (Death) minus 65 years, 5 months, and 4 days equals October 9, 1876 (Approximate Birth Date).

🎯 Offer Structured Next Steps

This document is excellent! It provides specific names and places to advance your research considerably.

Key research leads: The names of both parents, Frank W. Seaver and Harriet L. Hildreth, and the specific birth and death locations.

Would you like me to:

1. Analyze evidence quality (discussing the birth date/parentage information).
2. Suggest a search strategy for his birth record and the family in the census.
3. Help draft a citation for this document.

2)  Since it already offered structured next steps, I asked it:

Prompt 2:  What can you write about this now?

Response 2:  

Steve's Research Assistant v6.0

Custom Gem

This death certificate for Frederick W. Seaver, Sr. is a valuable foundational document for your genealogical research, connecting him directly to the previous generation.

Here is an analysis and interpretation of the document, focused on the Genealogical Proof Standard (GPS) principles of source and evidence quality:

📝 Analysis of Frederick W. Seaver, Sr.'s Death Record

Assessment: The certificate directly proves Frederick's death on March 13, 1942, in Lawrence, MA. The most critical new information are his parents: Frank W. Seaver and Harriet L. Hildreth. Since this information is secondary (based on recollection years later), finding a primary source (like Frederick's actual 1876 birth record in Leominster, MA) is the next vital step to confirm the parentage.

Key Genealogical Conclusions

Based on this single document, we can establish the following claims, noting the evidence status for each:

1. Death Confirmed: Frederick W. Seaver, Sr. died on March 13, 1942, in Lawrence, Massachusetts, and was buried in Evergreen Cemetery, Leominster, MA.
2. Approximate Birth: He was born approximately October 9, 1876 (65 years, 5 months, 4 days before death), likely in Leominster, MA.
3. Parental Leads: His father was Frank W. Seaver, born in Westminster, MA. His mother was Harriet L. Hildreth, born in Northboro, MA.
4. Residence: At the time of his death, he resided on Clinton Street in Salem, New Hampshire (N.H.). This is a crucial detail for finding him in the 1940 U.S. Census and other New Hampshire records.

Suggested Search Strategy (GPS Element 1)

The next steps should focus on confirming the secondary information and locating the family unit:

1. Locate Birth Record (Primary Goal): Search for a Massachusetts birth record for Frederick W. Seaver born in Leominster, MA, around October 1876. This should provide the primary evidence for his parents and exact birth date.
2. Locate Parents' Marriage/Death Records: Search for the marriage of Frank W. Seaver and Harriet L. Hildreth in Massachusetts, likely between 1870 and 1876.
3. Locate in U.S. Census Records: Find Frederick W. Seaver (age ≈ 63) in the 1940 U.S. Census (in Salem, NH, based on his residence). Then work backward to find him as a child with his parents (Frank and Harriet) in the 1880 and 1900 U.S. Census records in or near Leominster, MA.

Would you like me to focus on creating a complete citation for this document, or should we move on to the search strategy to find his birth record?

3)  I next requested a complete source citation:

Prompt 3:  Please provide a complete source citation for this document

Response 3:  

Steve's Research Assistant v6.0

Custom Gem

That is an excellent step. In genealogy, every piece of evidence must be accompanied by a complete, accurate citation (GPS Element 2) to allow for verification.

Since this is a certified copy of an original municipal record, a layered citation is appropriate, naming the original repository and the specific image you accessed.

Here is the complete source citation for the document you uploaded, using a standard genealogical format:

📜 Complete Source Citation

Full Citation (First Reference):

Massachusetts, City of Lawrence, "Certified Copy of a Record of Death, Frederick W. Seaver, Sr.," died 13 March 1942, recorded 18 March 1942, Death Records, Book 20, Page 100, No. 228, Office of the City Clerk, Lawrence, Massachusetts; digital image, FrederickWSeaver-1942-DeathCert-LawrenceMA.jpg, [User-provided file name], provided to Steve's Research Assistant v6.0.

Explanation of Citation Elements

4)  Finally, I asked for a strategy to find the birth record:

Prompt 4 was:  what strategy would you use to obtain a birth record?

Response 4 was:

Steve's Research Assistant v6.0

Custom Gem

That is the next crucial step in your research, as a birth record for Frederick W. Seaver will provide primary evidence to confirm the secondary information about his parents found on his death certificate.

Based on the information that Frederick was born in Leominster, Massachusetts, around October 9, 1876, the strategy for obtaining his birth record is highly focused.

🗺️ Strategy for Obtaining Frederick W. Seaver's Birth Record

The strategy uses the principles of Reasonably Exhaustive Research (GPS Element 1) and Jurisdictional Awareness for Massachusetts vital records.

Step 1: Determine Jurisdiction and Timeline

• Location: Leominster, Worcester County, Massachusetts.
• Time Period: 1876.
• Records: Massachusetts maintains state-level copies of vital records, but individual cities and towns also kept their own records. For this period (1876), records should exist at both the municipal and state level.

Step 2: Search the Easiest-to-Access State Records

The most efficient starting point is usually the state archives or the repositories that hold the digitized state copies.

1. Search FamilySearch and Ancestry (Indices/Microfilm): Massachusetts vital records (births, marriages, deaths) from 1841 to 1910 are well-indexed and often microfilmed or digitized.

• Action: Search the Massachusetts Births, 1841-1915 collection on FamilySearch or Ancestry using the name Frederick W. Seaver, the year 1876, and the town Leominster.
• Expected Result: A record or index entry confirming the exact date (October 9th) and providing the volume/page number, which is required to order the official copy. 

      2. Search the Massachusetts State Archives: The state archives hold the original                returns filed by the city/town clerks.

•  Action:  Consult the Massachusetts Archives Vital Records indices or collections for 1876.
• Expected Result: Confirmation of the record details.

Step 3: Search the Local Records (Leominster)

If the state-level index search fails, or to confirm the state copy, you can go to the municipal level.

1. Search Leominster Town/City Clerk Records: The original record would have been filed here first.

• Action: Search digitized Leominster Vital Records (often available on platforms like FamilySearch or local historical society websites) or check the published records of Leominster (sometimes found in genealogical libraries).

Step 4: Order the Official Record

Once you find the index entry, you can order a certified copy.

1. Ordering Source: Order the record from the Massachusetts Registry of Vital Records and Statistics or directly from the Leominster City Clerk.

• Action: Provide the specific name, date (October 9, 1876), and parents' names (Frank W. Seaver and Harriet L. Hildreth), and if found, the book and page number from the index.
• Reasoning: The record you receive will be a primary source (or a certified copy of the primary source) that definitively names his parents, which is the highest quality evidence needed to bridge the generations.

Would you like me to use the current information (Frank W. Seaver and Harriet L. Hildreth) to suggest a starting search for Frederick's record on FamilySearch?

5)  I stopped there. I could have asked for a search strategy for finding a marriage record, or census records, etc. 

6)  Steve's Research Assistant V6.0 Gemini Gem prompt works well.  

I have only a few quibbles:

• it used the term "primary evidence" in two places - one Response 2, and one in Response 4 - it should be "direct evidence."
• it used the term "primary source" once in the Response 2, and twice in the Step 4 discussion of Response 4 - it should be "original source" so that the user is not confused with "primary information."
• there is a "Primary Goal" term in Response 2.  I suggest "First Goal" or similar to avoid confusion.  

==============================================

Links to my blog posts about using Artificial Intelligence are on my Randy's AI and Genealogy page. Links to AI information and articles about Artificial Intelligence in Genealogy by other genealogists are on my AI and Genealogy Compendium page.

Copyright (c) 2025, Randall J. Seaver

The URL for this post is:  https://www.geneamusings.com/2025/11/u-saing-steves-research-assistant-v60.html

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post.  Share it on Twitter, Facebook, or Pinterest using the icons below.  Or contact me by email at randy.seaver@gmail.com.  Please note that all comments are moderated, and may not appear immediately.

Subscribe to receive a free daily email from Genea-Musings using www.Blogtrottr.com. 

RootsMagic Genealogy Family Tree Statistics Update - 1 January 2025

 I was curious to see how much progress I had made in my family tree during the last 12 months.

Here is the family tree database summary from RootsMagic 10 on 31 December 2024:

Here are the "numbers" from 31 December 2023 in my RootsMagic database (with increases from 1 January 2023):

*  73,962 persons (+ 2,155)
*  29.976 families  (+ 902)
*  231,476 events  (+ 6,240)
*  15,358 Alternate names (+ 336)
*  21,164 places (+ 2,016)  
*  2,286 sources ( + 59)
*  147,325 citations (+ 922)
*  1,265 Multi-media Items (+ 3)
*  3,211 Multi-media links (+ 2)
*  52,213 Persons matched to FamilySearch Family Tree persons (+ 2,001)

In the past 12 months, I've averaged adding 5.9 persons, 2.5 families, 17.0 events and 2.5 source citations each day.  I try to work at least an hour in the evening adding content and sources to the database, although baseball and the holidays intervened.

 Consequently, this year   managed to reduce my  citations per person from 2.04 to 1.99, and my  citations per event from 0.650 to 0.636 this past year.  I don't have a citation for every event, name or relationship, and in some cases I have more than one citation for an event, name or relationship.  Obviously, I have added more profiles and events with no or only a few source citations.

2)  I added 2,155 persons to my database in 2024 (+ 3.0%).  Some are in my 4th great-grandparents' descendant lines (including lines to DNA matches with known common ancestors), some are in my one-name studies, and some are from further and ongoing research on my ancestral families.  My one-name studies include Seaver (with variations Sever, Seever, Sevier and plurals), Carringer, Auble, Vaux, Dill, and Buck in my tree, and McKnew in my wife's tree.

By far, the most new research was done in finding the ancestry of my 4th great-grandmother, Mary (Partridge) Feather (1792-1853).    

3)  I "mine" and search new Ancestry, MyHeritage, Findmypast, and FamilySearch databases and use RootsMagic WebHints for my ancestral and one-name study surnames, and add content and source citations.  I add or correct name, relationship, date and place omissions or errors found while working on the database.  I use RootsMagic to match my tree profiles to the FamilySearch Family Tree profiles, and exchange source-verified information both ways on a near-daily basis.

4)  I stopped doing the 52 Ancestors in 52 Weeks challenge after 10 years in 2024.  I finished writing genealogical sketches for my ancestors from my parents to the 7th great-grandparents by ancestor chart numbers (although I skipped those for whom my research and/or reliable information is poor or non-existent).  

However, I decided to do a similar weekly challenge - 52 Relatives in 52 Weeks because the siblings of my ancestors deserve a genealogical sketch, and in most cases, don't have a sketch or life story in online collaborative trees.  The 52 Relatives posts are also in the 52 Ancestors in 52 Weeks page.  When I post a new sketch, I add the update the Notes and Sources to the person profile in RootsMagic and also to the FamilySearch Family Tree and WikiTree profile for that person.  I will continue this weekly challenge because it helps me focus on one specific relative, and their family, each week and improves my database and my family history.

5)  Another significant effort in 2024 was to convert current place names to historical place names as required.  As an example, Medfield, Norfolk County, Massachusetts, United States was Medfield, Suffolk County, Massachusetts Bay Colony, British Colonial America before 1776.  In the process, I'm taking the opportunity to refresh my Notes and Sources for my ancestors, and adding short biographicval sketches listing my "current thinking" to the ancestor's Note.  This is a big task and I'm about 3% done. 

6)  My conclusion is:  I've made steady progress, and I'm actively improving my database in both quantity and quality, but still have a long way to go to have a "fully sourced and accurate" family tree.  It's better than it was, but it can still be improved.  It is truly a lifelong task, I think!  I really need a genealogy clone or assistant.  I doubt that Siri, Alexa, Google Home or any other virtual assistant is yet capable of doing genealogy research, source citations, etc.;  I know that Artificial Intelligence is not able to do actual research, but it can help document submitted information.

 7)  I updated my Genea-Musings readers on the "numbers" in my RootsMagic genealogy family tree database over the years in:

*  RootsMagic Genealogy Database Statistics Update - 1 January 2024

*  RootsMagic Genealogy Database Statistics Update - 1 January 2023

*  My RootsMagic Genealogy Database Statistics Update - 1 January 2022

*  My RootsMagic Genealogy Database Statistics Update - 1 January 2021

*  My RootsMagic Genealogy Database Statistics Update - 31 December 2019

*  My RootsMagic Genealogy Database Statistics Update - 31 December 2018

*  My Genealogy Database Statistics Update - 1 January 2017

*   My Genealogy Database Statistics Update - 1 January 2016 

*  Genealogy Database Statistics Update - 1 January 2015 

*  Genealogy Database Statistics Update - 1 January 2014  

================================================

The URL for this post is:  https://www.geneamusings.com/2025/01/rootsmagic-genealogy-family-tree.html

Copyright (c) 2025, Randall J. Seaver

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post.  Share it on X, Facebook, or Pinterest using the icons below.  Or contact me by email at randy.seaver@gmail.com. Note that all comments are moderated, so they may not appear immediately.

Subscribe to receive a free daily email from Genea-Musings using www.Blogtrottr.com.