Glasshouse
Your organization's documents are unsearchable. Scanned PDFs, buried email attachments, SharePoint graveyards. Glasshouse extracts, indexes, and makes them all findable — with AI that actually reads the pages.
The Problem
Scanned Documents
Contracts, letters, forms — scanned as images. No search engine can read them. They might as well not exist.
Email Attachments
Critical documents distributed via email, buried in inboxes. When someone leaves, their attachments leave too.
SharePoint & Drives
The document graveyard. Files uploaded with good intentions, never found again. SharePoint search is notoriously poor.
What Glasshouse Does
Connect
SharePoint, email, network drives, cloud storage. We pull from wherever your documents live.
Extract
GPU-powered OCR reads scanned documents, tables, handwriting. Even low-quality faxes and photos.
Index
Dual search index: keyword (exact matches) and semantic (meaning-based). Find documents by what they say, not just how they say it.
Search & Ask
Natural language questions answered with citations to specific pages. "What was the outcome of the Henderson project?"
Proven: BC Government FOI Records
Glasshouse
These results return zero hits on the government's own search. The content exists only in scanned images that no search engine was reading.
What Glasshouse Is Not
- ✗ Not document backup — we index and search your documents, we don't replace where they live
- ✗ Not a primary repository — keep using SharePoint, Google Drive, your file server. We sit alongside
- ✗ Not a customer-facing website — this is an internal tool for your team to find information
- ✗ Not a generic chatbot — every answer is grounded in your actual documents with citations
- ✗ Not cloud-only — deploy on-prem if your data can't leave the building
Need something custom? We build bespoke solutions too. That's what we do.
Features
Multi-Tier OCR
Digital text, GPU OCR, table-aware extraction, cloud vision for the hardest pages. 98.9% extraction rate.
Dual Search
Keyword search (exact matches) and semantic search (meaning-based). "End of life care" finds MAID documents.
Conversational Q&A
Ask questions in plain English. Get grounded answers with citations to specific pages in specific documents.
On-Demand Extraction
Admins trigger cloud vision on specific pages with cost estimates upfront. Pay only for what you need.
Table Export
Structured tables extracted as HTML, downloadable as CSV. Turns scanned spreadsheets into usable data.
Cross-References
Semantically related documents surfaced automatically. See connections across departments and time periods.
Admin Dashboard
Coverage at a glance: pages extracted, methods used, cloud spend, per-department breakdown.
Bulk Ingest
Point at a source, pull everything in. Background processing with live progress. Handles hundreds of documents.
For Your Organization
Acme Corp — Document Search
Found a scanned handwritten inspection report from an email attachment. No other search tool would have found this.
Connect to Everything
Local Files
Network drives, mounted shares, local folders. Point and scan.
SharePoint
MS Graph API. Pull documents, metadata, and folder structure.
Exchange, M365, IMAP. Index bodies and extract attachments.
Google Drive
Google Workspace integration. Docs, Sheets, and uploaded files.
Cloud Storage
S3, Azure Blob, Dropbox, Box. Any object store with an API.
Your System
CRM, project management, legal case management. We build connectors.
How We Compare
| Glasshouse | MS Copilot | Elasticsearch | Glean | |
|---|---|---|---|---|
| Scanned document OCR | ✓ GPU-powered | ✗ | ✗ | ~ limited |
| Semantic search | ✓ | ✓ | ~ plugin | ✓ |
| On-premises option | ✓ | ✗ | ✓ | ✗ |
| Page-level citations | ✓ | ~ document | ✗ | ~ document |
| Table extraction to CSV | ✓ | ✗ | ✗ | ✗ |
| Setup complexity | Single deploy | SaaS | Complex | SaaS |
| Price (100 users) | Flat license | $5,700/mo | Self-hosted | Enterprise $$$ |
vs Microsoft Copilot
✓ We read scanned documents — Copilot can't
✓ On-premises option — Copilot is cloud-only
✓ Page-level citations — Copilot cites documents
✓ Flat license — Copilot is $57/user/month
vs Elasticsearch + Tika
✓ GPU-powered OCR for scanned docs
✓ Semantic search built in
✓ Single deploy — no Kubernetes/Java
✓ Conversational Q&A with citations
vs Glean
✓ Scanned document OCR — Glean is limited
✓ On-premises option — Glean is cloud-only
✓ Page-level citations
✓ Flat pricing — Glean is enterprise $$$
Deployment Options
Cloud Managed
- We host everything
- Connect your SharePoint / email
- Anthropic API for extraction
- Zero hardware
- Monthly subscription
Hybrid GPU
- Rented GPU (Hetzner / RunPod)
- Near-zero extraction cost
- Cloud API for hardest pages only
- Full data control
- We manage or you manage
On-Premises
- Your hardware, your building
- Nothing leaves your network
- Docker Compose deployment
- GPU required (any CUDA card)
- License + support contract
Let's Talk
We like it down here.