Highland Pro’s /lookup feature uses a combination of on-device and server intelligence to provide quick answers to simple questions while you’re writing.
Here’s how it works.
When you start a new line with a slash, Highland recognizes that you’re asking for something. For the rest of the line, everything you type is marked in gold. We call this the Request.
Once you hit return, Highland takes the Request — and only the Request, the text marked in gold— and decides what to do with it.
If the Request is something Highland can handle by itself on the device, it does so immediately. An example is /clear, which wipes away all previous /lookups and answers from the document.
<aside> <img src="/icons/arrow-right-basic_gray.svg" alt="/icons/arrow-right-basic_gray.svg" width="40px" />
We’ll be adding support for more on-device handling in future builds, including using Apple Intelligence for some Requests.
</aside>
If Highland can’t handle the Request on device, the app packages it up in a bundle consisting of two items: a JSON bundle with the Request as a text string, and a unique ID.
<aside> <img src="/icons/new-alert_orange.svg" alt="/icons/new-alert_orange.svg" width="40px" />
Highland’s servers never see your actual document, or anything beyond the golden words in your Request. If you’re curious, you can test this using a network packet monitoring app like Little Snitch.
</aside>
Highland’s server then looks at the Request packet and decides how to handle it.
If the Request begins with a known keyword — such as define, synonym or rhyme, the server finds an answer using either its internal database or an external database, such as Wikipedia.
The answer is then sent back to the user, with the answer printed in gold:
If the Request doesn’t begin with a known keyword, Highland’s server decides where to look for a good answer — but only if you have “Allow other lookups, including LLMs” turned on Settings. (If you have this turned off, Highland will tell you it can’t find an answer.)