Can Onyx handle legacy on-premises engineering data estate - PST, hard drives, network folders?

[aidan329]

We're evaluating enterprise search / RAG for a mid-sized engineering firm with approximately 20 TB of project data accumulated over 27 years. Sources include legacy hard drives, network folders, SharePoint, email archives and PST files. The data is messy — duplicates, inconsistent structures, scanned PDFs, old Office formats.
Core requirements: indexes in place without duplicating to cloud, permission-aware retrieval, citations back to source, LLM-assisted Q&A, on-premises preferred.
Can Onyx handle this, particularly the legacy on-premises file stores and PST archives? Native connectors or custom build required?
appreciate any thoughts/guidance.

[aidan329]

A few specific questions to make this easier to answer:
Old Outlook email files (PST): Can Onyx read these directly, or do we need to convert them to another format first?
Keeping data on-prem: I understand the original files stay where they are, but does Onyx always build its own separate copy (the search index) of everything? Or is there a way to avoid a second copy of the data?
Access permissions: If a file is restricted to certain staff, will Onyx respect that and only show it to people allowed to see it? And does that work for on-premises SharePoint and network folders, or only cloud sources?
Even partial answers would help. Thanks.

[Exhibitus-TroyThompson]

Following! (Are we at the same company?)

I built an in-house solution for local data indexing (Go/Typescript + ElasticSearch and a custom web UI) before discovering Onyx, and now find that Onyx can't natively read local data despite being installed internally.

Note to anyone considering Onyx self-hosting in its current state: the File connector is literally just files you upload, not a connection to any file system/file services. At least that's how the current Linux release behaves.