Silo indexes Windows file shares, Outlook, SharePoint, OneDrive and Teams, reads scanned PDFs and Rocscience models, and shows each person only what they can already open. It runs on your own hardware.
project:Ironbarkp.14… laboratory UCS results from 42 samples of fresh BIF average 86 MPa. The east wall design adopts 24 m benches at 65° face angle …
Materials: Fresh BIF (UCS 86 MPa, GSI 55, mi 17) · Weathered shale · Dolerite dyke. 3 stages, critical SRF 1.32.
Last 12 cores are in. Two east wall samples failed along bedding, so I've flagged them in the attached sheet …
Client wants the revised east wall FoS before the design review on Friday.
Plain words work. Filters narrow things down when you need them, and the filter list on the left updates as you go. These are the kinds of searches it was built for.
"east wall" UCS project:IronbarkThe 2024 wall review opened at page 14, the RS3 model, the lab results email and the Teams thread about rerunning section 4.
komatiite year:1995..2005Old reports scanned to PDF are read with OCR, so a 1998 annual report matches on page 37 like a modern Word file would.
type:rs3 material:BIFMaterial names, strength parameters and stages come out of Rocscience files, so nobody needs a licence just to find the right model.
commodity:gold has:locationCoordinates, MGA references, tenement IDs and map sheet codes are read out of the text. Draw an area to see what you hold there.
from:priya assays has:attachmentMessages, attachments and Teams posts sit in the same results as the documents they discuss.
project:2291A plugin reads project numbers and folders from your project system, so one search returns everything for a job, wherever it was saved.
Most search tools stop at Office files and PDFs. Silo handles those, then adds plugins for the formats geoscientists and engineers use. Each format is its own plugin, so new ones can be added without changing the rest of the system.
| Format | What Silo picks out | Status |
|---|---|---|
| Reports.pdf, scanned .pdf, .docx | Every page of text, with OCR for scans; titles, tenements, commodities, locations | Available |
| EmailOutlook, Exchange Online | Message text, people, dates and attachments | Available |
| Spreadsheets.xlsx, .csv | Sheet names, headers and cell text such as hole IDs and assays | Available |
| Rocscience RS3.rs3v3, compute, results | Materials and strength parameters, stages, supports, results | Available |
| Datamine.dm | Table type, fields, extents and coordinate ranges | In development |
| Teamschats and channels | Messages by project, channel and person | Enterprise |
Technical documents usually say where they're about somewhere in the text, even when the file name doesn't. Silo finds those references and puts the document on the map.
It understands MGA zones, AMG grids, 1:250k and 1:100k map sheets, graticular blocks and tenement numbers. You can combine an area with any other filter, and the built-in outline map works without an internet connection.
We test Silo against 46,850 public exploration reports from WAMEX and GSQ, many of them scanned. These are the numbers from a single workstation running the Silo Server beta. The organisation version spreads the same work across an index server and file-server agents, sized for shares of 70 TB.
Both use the same search. Desktop covers one person's PC and Microsoft 365 accounts. The organisation version covers every share, mailbox and channel in the business, with permissions.
Free for personal use during the beta. Indexes your local drives, the mapped drives you choose, and Outlook and OneDrive across several Microsoft 365 accounts. The index stays on your PC and nothing is sent to TekOne.
Installed on your own servers, one instance per firm. Indexes file shares of tens of terabytes, including deduplicated volumes, plus all of Microsoft 365, and checks access on every search. Licensed per site, with a pilot programme open now.
| Silo Desktop | Silo for organisations | |
|---|---|---|
| Who searches | You, on your PC | Everyone, from their PC or a browser |
| Local drives | Yes | Through Desktop |
| Network file shares | Mapped drives you pick | Whole servers, indexed on site |
| Outlook / Exchange Online | Your own accounts | Every mailbox, with access rules |
| SharePoint and Teams | OneDrive files | All sites, chats and channels |
| Technical file plugins | Yes | Yes, plus your own |
| Map search | Coming | Yes |
| Where the index lives | Your PC | Your server, on your network |
A small agent on each file server reads files and their permissions where they live, so nothing is copied across the network. The Silo server holds the index. People search from a small Windows app or a browser.
Silo reads the access rules from each file share, mailbox, site and channel, and checks them on every search. A contractor only finds what their account could already open, and HR folders stay with HR. Pick a person to see the same search through their account.
Yes. Silo Desktop is free for personal use while it is in beta. We'd like to hear what you think of it, and if it would help your whole team, we'll talk to you about Silo for organisations.
No. Silo builds its index on your own PC or server. Microsoft 365 content is read directly from Microsoft using your own sign-in, and stays on your hardware once indexed. Silo does not send your files, index or searches to us.
For file shares, the Silo agent reads the NTFS permissions on each file and folder. For Microsoft 365, Silo reads the sharing and membership of each mailbox, site and channel. It resolves users and groups from Active Directory and Entra ID and checks them on every search.
If anything about an item's access is unclear, Silo leaves it out of results.
PDFs (including scanned PDFs, read by OCR), Word, Excel, PowerPoint, email and attachments, text and CSV files, and Rocscience RS3 models. Datamine files are in development. Other formats can be added as plugins.
Silo for organisations is designed for file shares of 70 TB and more, with millions of files. Index size depends on how much text is in your data rather than raw terabytes: imagery, drawings and binary models add very little. Windows Data Deduplication volumes are supported.
One Windows Server (physical or virtual) at the site where your data lives, a small agent on each file server, and an Entra ID app registration for Microsoft 365. We help set it up as part of the pilot.
Most search tools were built for office documents and struggle with large technical archives. They index slowly, use a lot of server resources and can't look inside geoscience files. Silo was built for exactly that kind of data, and we're happy to run it side by side with what you have now.
No. We built it first with mining and geotechnical consultancies because their archives are some of the hardest to search, but it works for any firm on Windows and Microsoft 365.