Comparison of AI Platforms for Genealogy Reports

AI has undeniably become a valuable tool in genealogical research. Those not leveraging it risk missing key records and insights, while relying on it too heavily can lead to errors. Striking a balance is essential.

I consider AI my unpaid assistant (though I do pay for pro versions), allowing me to accomplish far more in less time. As genealogy is my after-hours passion alongside a full-time job, AI helps me generate, proof, and finalize research reports in record time—sometimes within an hour.

Many genealogists work solo, lacking someone to bounce ideas off. With AI, we now have a virtual research buddy.

Recently, I compared Perplexity and Claude with other WikiTreers. Some say they’re similar, with Perplexity acting as an interface for other models. For me, Perplexity consistently outperforms Claude in generating genealogical reports. Here’s the prompt I used for both, along with an ancestry profile and source list from Sourcerer:

Acting as a genealogist and using the genealogical proof standards, analyze these records and details. Provide a report and suggestions for research adhering to the GPS. Keep in mind that this family is part of a band of Romani Travellers, which will explain their multiple locations:

Claude

With Claude, I had to prompt a few rethinks. It struggled with the unique challenges of researching Romani Travellers, where typical records like land deeds and directories aren’t usually relevant. Romani Travellers often didn’t own land or have permanent residences, as their homes were wagons. Once I clarified this, Claude adjusted its research suggestions and ultimately provided a solid report.

Claude Repor:

Complete Analysis: Mathias Cooper (1846-1920) Romani Family

1. Current Identity Pattern Analysis

Name Variations Documented

• Mathias/Matthew/Mat/Matt/Mattie Cooper
• Mat Coffur (1920 census)

Context: Name variations are extremely common in Romani families due to:

• Oral tradition of name passing
• Deliberate variation when dealing with authorities
• Recording by officials unfamiliar with Romani names
• Use of different names in different contexts

Travel Pattern Evidence

1900: Pulaski, Beaver, Pennsylvania (House Dealer) 1910: Wade, Jasper, Illinois (Trader) 1920: Little Prairie, Pemiscot, Missouri

Children's birth locations create fuller picture:

• Tennessee (1876, 1885)
• Illinois (1885, 1887)
• Missouri (1888)
• Indiana (1884)
• Mississippi (1880)

2. Documentation Conflicts and Resolution Paths

Immigration Records (1859, 1868, 1900)

Likely Scenario: Multiple crossings rather than errors Research Path:

• Port city newspaper reports of Romani arrivals
• Police/court records in port cities
• Ship passenger lists focusing on group travel
• Border crossing records (particularly Canada)

Children's Birth Timeline

Two key patterns:

1. Multiple births in same year (1869, 1885)
2. Children born across multiple states

Research Approach:

• Newspaper coverage of Romani camps near birth locations
• Local court records for temporary camping permits
• Hospital records in major towns along travel routes
• Midwife records in rural areas
• Cross-reference with other Romani family births in same locations

Occupation Documentation

Listed as:

• House Dealer (1900)
• Trader (1910)

Research Strategy:

• Newspaper advertisements for horse trading
• Market day records
• Court records for trading licenses
• Police records of trading disputes
• Fair ground records
• Newspaper coverage of trading activities

3. Family Network Analysis

Marriage Connections

Marriage to Merinda/Miranda/Maranda/Maria Lock (1849-1934) 

Research Focus:

• Newspaper coverage of Romani weddings
• Court marriage licenses
• Other Lock family connections in Romani communities
• Newspaper mentions of Lock family trading activities

Parent Connection

Father: John "Fighting Jack" Cooper (1799-1878) Mother: Charlotte Clementina Lee (1796-1878)

Research Paths:

• UK transportation records (1840)
• Criminal court records
• Prison records
• Newspaper reports of "Fighting Jack"
• Australian penal records
• Tasmania colonial records

4. Priority Research Plan

Immediate Focus

1. Newspaper Research
   • Trading activities
   • Camp locations
   • Court appearances
   • Family gatherings
   • Horse fair coverage
2. Court Records
   • Trading licenses
   • Camping violations
   • Trading disputes
   • Marriage records
   • Minor infractions
3. Travel Pattern Documentation
   • Fair and market schedules
   • Seasonal movement patterns
   • Known Romani camping grounds
   • Traditional stopping points

Secondary Focus

1. Family Network
   • Other Cooper families in same areas
   • Lock family connections
   • Marriage patterns
   • Trading partnerships
2. Official Records
   • Hospital records
   • School records (temporary)
   • Vaccination records
   • Poor law records

5. Expected Record Patterns

Seasonal Documentation

• Spring/Summer: More records due to increased travel
• Winter: Possible permanent location records
• Market day documentation
• Fair ground records

Geographic Coverage

Focus on:

• Market towns
• County borders
• Traditional stopping places
• Major travel routes
• Fair locations

6. Research Challenges

Record Interpretation

• Official bias in documentation
• Deliberate name variations
• Missing or incomplete records
• Prejudicial reporting

Geographic Spread

• Multiple jurisdictions
• Various record-keeping practices
• Scattered documentation
• Inconsistent recording

7. Success Metrics

Primary Indicators

• Consistent family group identification
• Pattern of movement documentation
• Trading activity records
• Family network connections

Secondary Indicators

• Seasonal location patterns
• Inter-family relationships
• Trading route documentation
• Community connections

8. Next Steps

1. Immediate Actions
   • Map all known locations chronologically
   • Identify major market towns along routes
   • Locate newspaper archives for key areas
   • Identify court record repositories
2. Documentation Process
   • Create timeline of movements
   • Map family connections
   • Document trading patterns
   • Record name variations
3. Analysis Focus
• Movement patterns
• Family connections
• Trading activities
• Seasonal locations

Perplexity

Now let's compare that to the very same prompt given to Perplexity. Without any further instructions for edits, it provided this:

Genealogical Analysis Report: Mathias Cooper (1846-1920)

1. Reasonably Exhaustive Research The available records provide a good foundation for researching Mathias Cooper's life, but there are still gaps that require further investigation. We have census records, marriage records, and death records for some of his children. However, we lack direct birth and death records for Mathias himself.

2. Complete and Accurate Source Citations The provided information includes citations for various records, including census data, marriage indexes, and death certificates. These citations allow for the verification and evaluation of the sources used.

3. Analysis and Correlation of the Collected Information Mathias Cooper was born on January 1, 1846, in England. He married Merinda (also recorded as Miranda or Maranda) Lock on June 7, 1875, in Indiana, USA. The couple had numerous children between 1869 and 1904. Census records show Mathias (recorded as Mathew or Mat) in different locations:

• 1900: Pulaski, Beaver, Pennsylvania (age 54, House Dealer)
• 1910: Wade, Jasper, Illinois (age 61, Trader)
• 1920: Little Prairie, Pemiscot, Missouri (age 73)

Mathias died on March 18, 1920, and was buried in Marshall Cemetery, Clark County, Illinois.

4. Resolution of Conflicting EvidenceThere are some discrepancies in the records that need resolution:

1. Name variations: Mathias, Mathew, Mat, Matt, and Mattie are used interchangeably. This is common in historical records and likely refers to the same person.
2. Occupation: Listed as House Dealer (1900) and Trader (1910). These could be related occupations or indicate a change in profession.
3. Children's birth locations vary widely, which is consistent with the family's Romani Traveller background.
4. The 1920 census lists him as "Mat Coffur" instead of Cooper. This could be a transcription error or a temporary name change.

5. Soundly Reasoned, Coherently Written Conclusion Based on the available evidence, we can conclude that Mathias Cooper was a Romani Traveller born in England in 1846. He immigrated to the United States, married Merinda Lock in 1875, and had a large family. The family moved frequently across multiple states, which is consistent with their Traveller lifestyle. Mathias worked as a trader or dealer and passed away in 1920.

Suggestions for Further Research

1. Locate Mathias's birth record from England to confirm his exact birthplace and parents' names.
2. Search for immigration records to determine when and how Mathias arrived in the United States.
3. Look for additional census records between 1870 and 1900 to fill in the gaps in his residential history.
4. Investigate local newspapers in areas where the family lived for additional information about their activities and movements.
5. Seek out Mathias's death certificate to confirm the exact date and cause of death.
6. Research Romani Traveller communities in the areas where the family lived to provide context for their lifestyle and movements.
7. Attempt to locate any personal documents, such as letters or diaries, that might provide insight into Mathias's life and experiences.
8. Investigate land and property records in the various locations where Mathias lived to understand his "House Dealer" occupation better.

Conclusion  

Both AI platforms have their unique strengths and weaknesses. Claude’s output leans toward a structured outline, while Perplexity provides a mix of narrative and bullet points, which I prefer. However, the true measure of their utility lies in the accuracy and depth of the information provided.  

It’s critical to remember that no AI can replace human review of original documents. For example, both reports contained an error caused by a transcription mistake in Ancestry’s records. This error, revealed through Sourcerer’s source capture, misidentified the occupation as “House Trader” instead of “Horse Trader.” This distinction is significant, as “Horse Trader” was common among Romani Travellers during that era.  

When prompted about Romani Traveller life, Claude offered insightful research ideas, such as investigating horse fair coverage, trading licenses, camping violations, and trading disputes. These suggestions align well with the cultural context and offer promising avenues for further exploration.  
Perplexity, in contrast, produced a strong first draft that generally met genealogical proof standards but required additional details in its conclusion. However, neither platform thoroughly analyzed individual records for specific insights, leaving that task for manual review.  

It’s important to view AI-generated reports as starting points rather than final drafts. My process involves uploading original document images and asking the AI to compare the actual records against transcriptions, noting discrepancies, and extracting new information. This iterative process—refining prompts, editing reports, and adding details—results in a polished final document.  

There’s always room to improve how prompts are written to achieve better outcomes. I’m still experimenting with different approaches and would welcome ideas for creating effective genealogy prompts. What’s your favorite AI for genealogy, and why? Share your thoughts and tips!