Evidence first, relationships second
A conventional tree stores a relationship as a fact. AutoTreeGen stores the claim, the document it came from, the place inside that document, and what supports or challenges it. The relationship is the conclusion drawn from all of that, and it is allowed to stay undecided. This is why two trees that look identical elsewhere can look very different here.
The evidence graph
Open a tree and the evidence graph shows a person as the set of claims made about them, each with the source behind it and the confidence that source earns. It is the view to open when you want to know why the product believes something, not just what it believes.
This is the evidence behind a person - every claim with its source, its confidence, and the reasoning that accepted it.
Open:All trees
The evidence audit
Audit lists every claim the engines have made about the current tree, with its status on the five-step ladder, its confidence and the reasoning behind it. Pick a claim and the right-hand panel shows the evidence it rests on. Filter by status to see only what is still open. The page shows what the engines recorded and nothing else: where the record carries no source title or no DNA match, that field stays empty.
Open:Audit / Truth Engine
Sources and citations
A source is the document. A citation is the place inside that document a particular claim comes from - the page, the entry, the line. The distinction matters because a source can support one claim and say nothing about the next.
A source is where a fact comes from - a claim with no source is a lead, not a fact.
The sources library
Citations is the library of the documents behind your tree: one row per source, with its type and the claims it is cited for. Open a source to see where it came from and what it supports. A source's overall quality is not scored here; quality lives on each citation, because the same document can be strong for one claim and silent on the next.
Open:Citations
Conflicts
Two birth dates, two places, two sets of parents, a family memory against a document, a modern transcription against the original register - each of these is a conflict, and the tree carries them as conflicts rather than picking a winner quietly. Some conflicts stay open for years. An open conflict is a legitimate state, not a defect in your tree.
Hypotheses
A hypothesis is where a research idea lives while it is still an idea: could these two families share a father, could these two spellings be one man. Several competing hypotheses about the same question can be kept alive at once, each with its own reasoning and its own evidence, and none of them touches the tree until you decide it should.
A hypothesis is a possible answer we are not yet entitled to call a fact.
Open:Hypotheses
Research decisions
When you accept, reject or postpone something, that is itself recorded: what the claim was, where it came from, and what was decided about it. A rejection is not an erasure - it is a decision with a reason attached, which is what stops the same rejected link from quietly reappearing through the next import.
Deciding on many hints and hypotheses quickly
On Tree hints, each row has Same person and Not the same buttons. On the hypotheses list, each row has Accept and Reject. A decided row leaves the list at once, and a message offers Undo. To decide many at a time, tick the rows, or select every row on the page, and use the bar at the bottom: confirm, reject or decide later for all of them. You can also use the keyboard: the arrow keys or J and K move between rows, Y says yes, N says no, L means later, X ticks a row and Enter opens it. If a save fails, the row comes back with the reason and a way to try again.
Why a document counts more than a copied tree
How sure we are of a relationship depends on what kind of sources stand behind it, not on how many there are. A birth, marriage or death record counts the most, an official record such as a census counts a lot, and someone's family tree counts little. All family trees together count as one piece of evidence, because five copies of one tree are still one tree. A relationship is shown as confirmed only when two independent documents support it; family trees alone never confirm it. In the evidence panel each source carries a badge with its kind: Vital record, Official record or Someone's family tree.
Links without documents and line strength
Links without documents lists the families in your tree where no birth record or other document proves that the children belong to these parents: they rest only on copied family trees, or on nothing. That is your list of what to look for in the archives. A large tree is checked part by part; press Check more links to go on. On a person's page, How well this line is documented shows each generation up the line: how many parent links are backed by a document, how many only by copied trees, and how many have no source.
Leads for a brick wall
When you mark a person as a brick wall, their page shows Leads for this brick wall. The platform searches for the surname, with its spellings and the forms that sound alike, in three places: your other trees, trees that other members share, and your DNA matches, where it looks in the match name, your group label and the common ancestor the DNA site named. Each lead says why it was found, for example the same surname, a spelling of it, or the same birthplace, and links to the person or match. Names from shared trees open on a plan with community search; on other plans you see how many there are. If a place was not searched, the panel says why.
Community evidence index
Opting in compares the deceased people in your tree with those in other members' trees - by name, years and place, every night at 02:20 UTC. You see how many of your people appear in someone else's research; on Pro and Premium each one arrives as a match you review, and nothing is ever merged. Living people, DNA and anything you have not opted in never leave your tree.
Open:Shared trees
Provenance
Every record in a tree carries where it came from: which file it was imported from, which job brought it in, and which edits were made by hand afterwards. When two people in your tree are merged, the merge keeps a log of what came from where, so the join can be read back later.
A person's workspace
A person's page shows their names, events, relatives and sources on one screen. The evidence workspace behind it lists every claim about that person with the evidence on each side, and Line draws one ancestor's descent with the DNA that backs each step. When two records turn out to be one person, merge them from the duplicates page; the merge log on the surviving person records what came from where, and a merge can be undone for ninety days.
Collection: the lines inside one tree
Collection groups the people of a tree into the lines they belong to, named by the family they descend from, and shows where those lines touch: a marriage, a shared ancestor, a DNA bridge. It is a map of the tree, not a verdict on it. No line is labelled with a nation or an origin, and a tree whose lines are not yet computed shows that it is empty for that reason, not an invented map.
Calendar converter
Dates in old records come in more than one calendar. The converter turns a Gregorian, Julian or Hebrew date into the others, including leap months, and says whether the conversion is reversible. A date that cannot be converted back to itself is marked to treat with caution rather than returned as if it were certain. The tool converts; it does not change any date in your tree.
Open:Evidence calendar
Research reports
A report gathers the current conclusion on a question together with the evidence behind it, the conflicts it has to address and the questions still open. It is written to be read by another researcher, and it is regenerated from the graph rather than typed once and left behind.
A report is the proof behind a conclusion - not a pretty narrative.
Open:Reports
Research and tools overviews
Research shows how many hypotheses are open on the current tree, how many wait for your decision and which engines have looked at it. Tools lists every tool with the date it last ran and what it found then. The tree's Engines page names the eight reasoning engines and, for each, whether it ran on this tree, did not run, or has never reported. The Evidence Report gathers every engine's findings into one printable document. Each of these pages reads live data, and a tool that never ran is shown as never run, not as having found nothing.
Open:Research
Evidence certificates
A certificate is a fixed, content-addressed snapshot of an evidence bundle: the same inputs always produce the same certificate, so a conclusion you shared last year can be checked against the evidence it actually rested on.
Open:Evidence certificate
Need help with this step?Ask the assistant