How we work

There is one reference shoot, and it is the source of every count we publish. It is a real wedding, shot by the person who wrote the software, not a benchmark set assembled to flatter it. Same card, same machine, network off, running on battery so nothing is hiding behind a wall socket. The clock starts when the card is mounted and stops when the last frame has a verdict.

That it knows which photograph is good. It does not, and it is not trying to. It clears what is plainly unusable and names why. Ranking your favourites is not a feature we are behind on, it is one we decided against. That it has learned your taste. There is no profile of you, no training on your selects, no model of what you like. If that ever changes it will be something you switch on, and this page will say so before the marketing does. That the numbers above generalise. One wedding is one wedding. We have not published a multi shoot study, and until we do, the honest claim is "this is what it did here", not "this is what it does". That it is finished. Kepla for Mac is in private preview. Founding Access places are limited and the software is still changing under the people using it. The bar it has to clear before any of that softens: 90% agreement with a hand cull as a floor, 95% as the target, and zero false rejects on hero frames. A hero thrown away is the only failure that cannot be undone by a photographer who trusts the tool.

Being straight about this, because the dates on this blog make the pattern visible anyway and we would rather explain it than let you infer it. The guides on this site are drafted with AI assistance, from an outline and a source list assembled first, and then reviewed before publishing. A large batch went out in August 2026, which is why so many posts share a publish date. Nothing here is published unreviewed, and nothing is published to fill a keyword slot that we had no answer for. What that process is held to: every post cites its sources so you can check the claim rather than trust the byline, including the vendors we compete with. Competitor pricing and features are taken from the vendor's own pages. Where a rival tool is the better buy, the comparison says so. Where Kepla has not been tested against something, the post does not pretend otherwise. The judgement, the corrections and the responsibility are one person's. If a post is wrong, that is on him, not on a tool.

Competitor prices move without warning and a stale price is the most likely thing on this site to be wrong. If you find one, or any other factual error, send it to support@getkepla.com and it gets fixed. Modified dates mean something here. A post's modified date moves only when the content actually changed. We do not bump dates to look fresh, which is why most posts on this site still carry their original date and will keep carrying it until there is a real reason to change. What gets reviewed, and how often. Posts carrying prices, comparisons and product claims are reviewed on a quarterly cycle. Posts that lean on a competitor's specific feature or a specific Adobe behaviour are reviewed twice a year. Craft guides are left alone unless something has drifted. If a number on this site cannot be traced back to the wedding at the top of this page, it should not be there. Tell us and it comes down.

A wrong price, a claim that does not hold, a frame Kepla cleared that it had no business clearing. All of it is worth an email.