Keeping the source on every claim
When an answer is built from many places, the risk is that the source behind each claim quietly falls away as findings are tidied into one confident summary. Once it has gone, a checked figure and an offhand guess read exactly the same. The wider set of decisions this sits inside is Prompt engineering with Claude.
When an answer is built from many sources, a research task, a set of documents, a handful of tools, the last question is whether you can still trust where each part came from. The risk is that as findings are gathered and boiled down into one tidy answer, the source behind each claim quietly falls away. Once it has gone you can no longer tell a checked, authoritative figure from an offhand guess, because both now read as the same plain sentence. Keeping the source attached to each claim is what lets the final answer be weighed rather than simply believed.
Two things follow from keeping the source. When several sources independently say the same thing, that agreement is itself a signal, so a claim backed three ways is stronger than the same claim said once, and combining findings should keep all three rather than tidy them down to one. When sources instead disagree, the move is not to quietly pick one and carry on, which manufactures a false certainty and hides that there was ever a question. You show both with their sources and send the conflict to a person, because choosing between credible sources needs context the AI does not have.
Dates matter for the same reason. Two figures that differ are not always in conflict, they may be the same number measured at different times, and only the date tells them apart. A price set on this month’s conditions is rightly different from last year’s, not a contradiction. Requiring the date alongside each claim is what stops a normal change over time being read as a clash.
Finally, how the answer is laid out carries meaning of its own. Findings everyone agrees on and findings still contested do not belong in the same flat list, where the shaky reads as solid as the sure. Separating them, and giving figures a table where the rows can be compared while narrative stays as prose, lets the reader see at a glance what is well-supported and what is thin.
Say it another way
Think of a recommendation reaching you two ways. The waiter who has known you for years and knows your taste says the fish is the one to have. A glossy advert on the television says the same about a meal. Both reach you as “this is good”, but they are not the same signal at all, and if you forget who each one came from you would weigh the advert as heavily as the friend who knows you. Keeping track of who said a thing, and when, is what lets you trust it the right amount.
Doing this with Claude Code
When Claude gathers an answer from several places, ask it to carry the source of each part through to the end rather than dropping it in the summary.
Keep each claim with its source. Have the model attach where a claim came from and when, and carry that through when it summarises, so a checked figure is never flattened into the same plain sentence as a guess.
Treat agreement as strength. When several sources say the same thing, keep all of them rather than reducing to one, because the agreement is part of why the claim can be trusted.
Show conflicts, do not pick. When credible sources disagree, have it surface both with their sources and route the choice to a person, rather than quietly settling on one the AI has no basis to choose.
Ask for dates, and lay it out by confidence. Require a date on each claim so a change over time is not read as a contradiction, and separate the well-supported findings from the contested ones so the layout itself shows what is solid.
Back to Prompt engineering with Claude. The craft around it is Working with Claude.