A Soldier's Life

Every war, every record, one honest map

Which records exist for your ancestor depends on his war and his side, and where a cell below says no federal record exists, that's worth knowing too. Prices link to each war's own page.

A record's reliability is not uniform across these wars, and the report never pretends it is. A Revolutionary War pension rests on sworn memory given fifty years after the fact; a Civil War muster roll was written the day it happened. We weigh each by how close it stands to the event, and say so on the page.

Ordering two or more soldiers? Ten percent off the total. And historian-register project work (a regiment's worth of pension files, say) is quoted as a project, not priced by the row. Write first; we'll scope it together.

A report you can act on

Written for a family reader, not a research audience. Every finding points somewhere you can go next.

• Names, dates and places, pulled out and listedEvery person, residence, occupation and date in the file, set out as leads, each one pointing at a census, a county record, a cemetery.
• What the file establishes, and what it doesn'tIf the record proves his service but not the link to your third generation, we say so and tell you what would.
• Every hard-to-read entry listed honestlyWhere a word can't be read, it's flagged. Nothing is guessed at, and no gap is quietly filled in.
• The law behind each claimA veteran who re-filed in 1890 with no new wound was usually following a new law, not failing health. We tell you which changes are the man and which are the statute.

Read one before you order

Complete reports, start to finish, including the parts where the record wouldn't answer the question. Some sample soldiers are invented so that no real family's file is published; the rest are clearly marked "real file": public records like a celebrated 1864 court-martial, Confederate pensions the state archives publish freely, and the founder's own research on his camp's namesake. Every document type, tier, flag and citation is exactly what you'd receive.

All the sample reports, grouped by war: one for every product →

Why not just run it through AI yourself?

Fair question, and you genuinely can. What you'll get back is a decent summary of the pages you handed it. What you're buying here was never the software: it's the researcher the method was built from. Three of the four reasons below come from a lifetime in these records; the fourth comes from the economics of the tools themselves.

• Knowing what should be thereA file can only be summarized; a claim has to be known. An 1818 pension should carry an 1820 property schedule. A veteran's Confederate file was never built to hold his death record, and the report tells you which record was. Our reports name what's missing from the file and where it actually lives, because the method was written by someone who has pulled thousands of these. A tool answers the file it's given. A researcher knows the file it wasn't.
• Reading witnesses like a historianThe army recorded one death date; the widow swore to another, because she was told of his death, not there for it. Every statement in these files was made by someone, for a reason, under some incentive, and our reports weigh each one by who created it and why: the disagreement stays on the record, with our read and the reason stated in the open. A summary smooths this over. Source criticism refuses to.
• Knowing when to stop and lookSome files hold an ink-blotted bounty no honest reader can decode, or two same-named soldiers no index can safely tell apart. Our reports flag those moments by page and refuse to guess. The software won't ship a report unless every page is accounted for, and the reviewed tier puts a person back over every flagged entry against the original scans. Knowing where the hard cases are is the researcher's half of the job.
• The free-account realityOur experiment was the generous case: a paid research account reading all fifteen pages in a single pass. The free chatbot subscription most families would actually use meets a pension file differently: limits on how many pages you can attach, message caps that arrive mid-file, and quiet downgrades to lighter models. So the file goes in as fragments, and a model reading fragments cannot do the one thing this work depends on: holding page 3 and page 14 in view at once to notice they disagree. A 100-page contested widow's claim never fits at all. By the time you have bought the access and built the process to feed a whole file honestly, you have spent more than the report costs, before the first page is read.

What you are paying for, plainly. A researcher-designed reading of the complete file, with every fact tied to the original page that states it. If you're building a lineage-society application, that structure is the point: societies accept the original records as evidence, and this report is the finding aid that walks you to them, page by page, ready to verify.

And plainly about the tool. A person designed this method and the software drafts within it, never outside it. It will not ship a report unless every page is accounted for, and the reviewed edition puts a researcher back over every flagged entry against the original scans. The AI is the draftsman here, not the historian.

Grounded in the statutes themselves, read in full

A pension file is a stack of forms answering a law most families never see. Every report names that law: the acts each claim was filed and decided under, with a plain synopsis of what each one required, tied to the document in your file that turns on it. When a claim is approved, cut, or refused, you learn the statute that caused it, not a guess. When a rate suddenly jumps, you learn the act that raised it.

• The actual acts, not a secondhand summaryThis rests on the pension statutes as Congress and the state legislatures actually wrote them, read in full from the original session laws. Not a textbook paraphrase, and not a chatbot's half-memory of a law it never opened. Where a governing act cannot be confirmed, the report says the basis is unverified rather than inventing one.
• Which changes are the man, and which are the lawA Union pension that leaps in 1907 is usually the Service and Age Act, not a worsening wound. A Confederate widow's claim, refused one year and allowed the next, often means the marriage-date line moved, not that she found new proof. The report separates the life from the legislation, so you read the file correctly.
• Federal and state, across every war we coverThe federal pension laws of six wars, from the Revolution and the War of 1812 through the Mexican War, the Indian Wars, the Civil War, and the War with Spain. And on the Confederate side, the separate pension systems of all fourteen states that built one, from Texas to Virginia to the border states, each read for its own classes, rates, residence rules, and widow tests, state by state and era by era.

Why this is rare. No plain AI run carries it, and few reports of any kind do, because it takes reading the laws themselves, act by act, before a single file is opened. It is the difference between telling you what a form says and telling you the rule the veteran or his widow actually faced, and why the government answered the way it did.

What this isn't

What a report costs

Prices are on the coverage map above and on each war's own page, where each report is described in full. You send the file you already have, and if you don't have it yet, here is exactly how to get it.

Automated edition. Every report at these prices is a finished automated report. We read it through once before it goes out to catch obvious errors, but we don't check each reading against the original documents, and the report tells you so, every time. Reviewed edition. A person re-reading every flagged entry against the original scans, opening later. Join the waitlist and today's prices are held for you, whatever they become.

If it isn't your soldier, you don't pay

Same-named soldiers are the oldest trap in this work, and even careful families sometimes pull the wrong man's file. So the guarantee is simple:

• Wrong soldier: full refund, and we read the right file free. If the report comes back and the man in the file isn't your ancestor, you get your money back. Then we help you find the correct file, and when you have it, we read it at no charge. That mistake costs you nothing but the ordering fee the archive charged, and you end up with the report you actually wanted.
• Not satisfied: tell us why, we re-run it, then refund. If the report disappoints you, write us with the reasons. We build the report again with your feedback in hand, once, free. If the second reading still doesn't satisfy, full refund. No fees, no fine print, no restocking charge on a family's history.

We stand behind this because we're confident in the process: every order is checked against the soldier you named before it ships, and every report states plainly what the file establishes and what it doesn't. What matters to us is that you end up satisfied, holding the most accurate account of your soldier the records can give.

One honest note on the two editions. The Automated report traces every fact to its page in a single complete pass, and tells you on the page wherever a hand should still check the original. The Reviewed edition is that second hand, a researcher going back over every flagged entry against the scans. The guarantee covers whether it is your soldier either way; the editions are how far you want the reading pushed on everything around him.

Who reads your file

Sean Protas

Historian · Author · Sons of Union Veterans

I've read more than two hundred pension files researching the 2nd Delaware Infantry. That's where this came from: enough files to know what these records reliably establish, and what they only appear to.

The hard part was never the handwriting. It's knowing that a pension rate rose in 1907 because Congress changed the law and not because a man's health failed, that an R on a Revolutionary War file doesn't always mean the claim was rejected, and that a widow's sworn account of a marriage forty years earlier is strong evidence of one thing and weak evidence of another. Those distinctions are what you're paying for.

In the field with a reenacting unit since 2005, including a few turns as a soldier on film. A history degree from the University of Delaware.

My own family lineage runs to Musician Charles H. Minnemeyer of the 11th Pennsylvania Reserves Infantry, the line my Sons of Union Veterans membership is documented on. I've had to prove a lineage myself, out of these same records.

The deepest archival mileage I have is in Civil War Union files, and that is where this method was built and proven. What carries it into the other five wars is not a claim to equal depth in each. It is the method itself, the same tiered reading of the evidence, and the fact that I went to the statutes of every war and every Confederate state and read the pension acts from the original session laws before a file was ever read against them. The expertise that travels is the discipline. The law is read, not remembered.

I built the process that produces these reports, and everything about the way it flags uncertainty rather than filling it in comes from knowing what these tools do when nobody constrains them. I'm a historian and researcher, not a board-certified genealogist. That is a real credential, and it is not one I hold. What I bring is documentary knowledge: I know these record groups, the acts behind them, and the places where they go quiet.