The FoxTrot alternative for when you can't remember the words
FoxTrot and Loupe made the same unusual decision: build an index rather than ask Apple's. What they put in it is where they part company, and it is a real choice rather than a better-and-worse.
FoxTrot indexes text out of a very long list of file formats and matches it literally, with proximity, boolean and regular-expression queries on top. Loupe covers fewer formats, but reads pictures as well as text and keeps a second index built on meaning, so a document can match a query that shares none of its words. Breadth against depth, and which one you need depends entirely on why your last search failed.
The thing they have in common
Almost every Mac search tool is a front-end to Spotlight. Apple's index is already there, already populated, and already knows hundreds of attributes about every file, so building on it is the sensible engineering decision and most products make it.
FoxTrot did not, and neither did this. Both open the files themselves and keep their own index, which costs an initial read of your disk and buys independence from whatever Apple decided to index this year. If you are comparing these two rather than comparing either to Spotlight, you have already worked out why that matters.
Where they part company
FoxTrot's answer to "what should be in the index" is: the text of as many formats as possible, from as many places as possible, with the sharpest possible query language on top. Decades of formats, network volumes, millions of files, proximity operators, regular expressions, several separate indices, and a server so a team can share one.
The answer here is narrower and deeper. Fewer formats, one Mac, no server, and a deliberately plain query box. What goes in instead is everything that was never text to begin with: scanned pages read by character recognition, photographs described by a vision model, and a second index built on meaning rather than spelling.
Neither is the better answer in the abstract. They fail at different searches.
| What you're trying to do | FoxTrot | Loupe |
|---|---|---|
| Index a very wide range of file formats | Its headline strength. A far longer format list than anything here. | Documents, spreadsheets, presentations, iWork, images, plain text and source. |
| Search without relying on Spotlight | Builds its own index. The reason it can reach what Spotlight will not. | Also builds its own, in a single file, and never touches Apple’s. |
| Find two words near each other, or match a pattern | Proximity, boolean and regular expressions, with hits highlighted in colour. | Simple filters. Nothing like this. |
| Find “the lease” when the document says “tenancy agreement” | No match. The index matches the words you typed. | A meaning index alongside the keyword one, so the document still comes up. |
| Find the photo of a whiteboard from a workshop | Not what it is for. A picture with no text in it is a filename. | Describes the scene with a vision model and indexes the description. |
| Let a team search a shared archive | A search server with per-user access, sold for exactly this. | Nothing. One index, one Mac, no accounts. |
| Search network volumes and archives | Built for it, and scales to millions of files. | Local folders you tick. Not designed for a file server. |
| Run it on an Intel Mac, or an older macOS | Both, with legacy builds going a long way back. | Apple silicon and macOS 26 or later, full stop. |
Which search are you failing?
The useful question is not which tool is stronger but what happened the last time you could not find something.
You knew roughly what the document said, and could not narrow it down. That is a precision problem, and precision is FoxTrot's entire case. Proximity search in particular — two words within a few of each other — is a genuinely powerful thing that nothing here replaces.
You could not remember any of the words. You remembered that it was about the boiler, or that it was the thing the landlord sent. No query language helps with that, however expressive, because you have nothing to put in it. A meaning index is the only one of the two that can answer a question phrased the way people actually remember things. The same argument, made against Spotlight.
The thing you wanted is a picture. A photograph of a whiteboard or a receipt on a kitchen table contains no text to index, so a text index of any breadth has nothing to store. How that gets read here.
Where FoxTrot is the right buy
If your archive spans thirty years of formats, lives on a file server, has to be searchable by more than one person, or has to run on an Intel Mac, this is not a close call and you should buy theirs. Those are real requirements and none of them are on this product's roadmap.
Loupe is for one person, one Mac, and the problem of not remembering. That is a smaller product on purpose.
Questions people ask about this
Does FoxTrot use the Spotlight index?
No, and that is the most interesting thing about it. It builds an index of its own, which is why it can read formats Spotlight has no importer for and search volumes Spotlight will not touch. It is the only other Mac search tool in common use that made that choice.
Then what is different here?
What goes into the index. FoxTrot indexes text out of a very long list of formats and matches it literally, with proximity, boolean and regular expressions on top. Loupe covers fewer formats, but also runs character recognition over scans, has a vision model describe pictures that contain no text, and keeps a second index built on meaning so a query can match a document that never uses its words.
Which one handles more file types?
FoxTrot, by a wide margin, and it is not close. If your archive contains formats from the last thirty years, that breadth is the thing to buy.
I need several people to search a shared archive.
Then this is not the tool, and FoxTrot might be. They sell a search server with per-user access, which is a category of product that does not exist here. Loupe is one index on one Mac.
I have an Intel Mac.
FoxTrot runs on Intel and on macOS versions well before this one. Loupe needs Apple silicon and macOS 26 or later, because the models that read your pages run on the GPU.
Try it on the search that failed.
Think of the thing you gave up looking for, and describe it rather than spell it. Seven days, no card, and nothing of yours leaves the machine.
- Apple silicon · macOS 26 or later
- No account
- Nothing leaves your Mac