Is there a reason to believe this is something enterprises actually want? Was GSE too early?
Whether or not you can actually make enough money from building a selling on-prem search is another question. It is an incredibly difficult space to develop for, because every one of these messes has a vastly different IT stack. [1][2][3]
You could take the easy way out and build search for Slack + Jira + Email, but that would help less than 1% of businesses world-wide.
[1] The first full-time job I worked had an unholy amalgamation of HP Quality Center, SeaPine source control, something about Sharepoint, and random documents stored in people's shared folders. Oh, yeah, and it had an on-premise Google Enterprise Search (Or whatever it was called back in 2011), which was almost entirely useless.
[2] The second was a mix of Jira and TikiWiki.
[3] The current job my wife holds has a bunch of files in a shared directory, a database that is installed from four floppy disks, and an IT stack that quite prominently includes Windows 2003.
I have a strong feeling that Google Desktop Search had a lot of this sorted.
That was 10 years ago.
With all the advances that has been happening in the meantime and assuming businesses are interested enough to adjust a tiny bit to simplify crawling this doesn't seem impossible at all if a well-funded team started working on it.
You can't actually leverage any of the economies of scale that you can get from building software, when you have to hold the hand of every single customer you deal with.
Someone like Oracle[1] can get away with this sort of thing (While charging an arm and a leg for it), but this is not the problem that your typical startup can solve.
[1] The other difference is that your Oracle database + solution is business critical, and businesses have no problem with paying $XY,000/year for business-critical solutions. Intranet search is a nice-to-have, and businesses balk at paying $5/user/month for that sort of stuff.
Slack search is awful and they have a full team on it. Jira is a mess. Gmail search is pretty good already, which is one of the things that makes frontapp far less useful also.
Getting everyone to put everything in confluence - for example - helps, but... there are entire categories of data that don't end up getting stored in confluence, so... it wouldn't get found. And, fwiw, the confluence search always seems to be not very helpful whenever I'm needing to find something - but I often can't tell up front if the search is bad or the data just isn't there (or I'm searching for the wrong terms).
If they don't want it, though, they won't buy it, whether it's helpful to people on the front-lines or not.