PRODUCTS
Intelligent Deep Storage™ for
Microsoft Azure
Enterprise file storage that tiers inactive data into cost-optimized object storage, protects it, and keeps every file searchable. Deployed entirely inside your own Microsoft Azure subscription.
CAEVES combines the economics of cloud object storage, the usability of modern file systems, and the requirements of AI-ready data access into a single platform that operates within the customer’s own Microsoft Azure tenant.
CAEVES Intelligent Deep Storage™ and the Copilot Connector™ ensure that discovery and AI-driven reasoning provides end-to-end security and authorization.
As a result, data flows seamlessly from file system to object storage to search and AI, without duplicating repositories or introducing separate archive workflows.
THE PLATFORM
Two phases. One data foundation.
Intelligent Deep Storage™
CAEVES Intelligent Deep Storage™ delivers a high-performance software-defined storage platform, allowing organizations to reduce storage and data management TCO through data caching, tiering, and lifecycle management policies, while preserving multiprotocol file access, continuous accessibility, security, and data governance.
Copilot Connector™
The CAEVES Copilot Connector™ indexes Intelligent Deep Storage™ stored data and brings AI to your private data sets, securely within your own (hybrid) cloud environment.
Get instant, traceable answers from your historical data using Microsoft 365 Search & Copilot, Teams, or other Microsoft applications, always referencing to the source origin files.
Platform Highlights
Multiprotocol File Access (SMB/NFS) providing real-time data access
Users connect to a high-performance, NTFS-compliant virtual file share over SMB, with NFS for Linux and cross-platform clients for data ingest, migration and retrieval.
Caching & Data Tiering from Premium SSD to Azure Object Storage
Active data is cached locally on Premium SSD for full performance, while inactive file contents are stubbed and tiered to object storage. Metadata stays on the instance and payloads move to object storage, giving you object-storage economics without sacrificing file-system responsiveness.
Automatic Data Tiering to Hot, Cool, Cold and Archive blob access tiers
A policy engine built on Azure Data Lifecycle Management classifies file activity and moves content across Hot, Cool, Cold, and Archive tiers automatically. Every threshold is configurable, so retention follows your project lifecycles.
Public, Hybrid & Edge Cloud Deployment
CAEVES runs entirely inside your own Azure environment in whichever topology fits: public cloud with instance and storage in your tenant, hybrid with the instance on-premises and object storage in Azure, or edge-to-cloud with distributed agents feeding a centralized or regional Azure storage account.
Copilot Connector™ securely indexes your privately managed data to M365
The Copilot Connector™ indexes your stored objects into Microsoft Graph, in place, with inherited ACLs intact. Decades of archived files become answerable through Microsoft 365 Search and Copilot, with no duplicate store and no custom retrieval pipeline to build.
Create role-specific agents, integrate with Copilot or Copilot Studio
Deploy dedicated Copilot Studio agents with their own data scopes, personas, and governance instructions. Each agent reasons over CAEVES-indexed content within the user's own permissions, so different teams get purpose-built assistants grounded in the same governed archive.
Multilayer Security model with Active Directory and Entra ID federation
Access rights are defined with standard NTFS ACLs at ingest and federated through Active Directory or Microsoft Entra ID. Permissions are enforced at every tier, embedded in the search index, and trimmed by Microsoft Graph at query time, so users only ever see what they are authorized to access.
FUNDAMENTALS - INTELLIGENT DEEP STORAGE™
Built for Microsoft Azure
CAEVES presents a high-performance, NTFS-compliant file share and transparently tiers inactive file data to Azure Blob storage. Metadata stays local and instantly browsable. Contents live as standard objects in your storage account. The CAEVES tenant holds only capacity metering data. No file names, no metadata, no content.
Figure 1. CAEVES reference architecture. Every component except Microsoft Graph runs inside the customer's own Azure subscription, in the customer-selected region. NTFS ACLs follow the file from ingest through object storage into the search index.
| Component | Azure service | Role |
|---|---|---|
| Software Subscription | Microsoft Marketplace Offer | Subscription activation, capacity metering, billing. First 5 TB free, excluding Microsoft Azure infrastructure cost. |
| CAEVES Instance | Windows Server Virtual Machine | Hardened virtual appliance. SMB / NFS file share, caching, tiering engine, snapshot orchestration, migration engine. |
| Object Storage | Azure Blob Storage | Authoritative payload store for all tiered file data and snapshots. Hot, Cool, Cold, and Archive access tiers. |
| Metadata | Azure Table Storage | File-to-object mapping, NTFS security descriptors, connector process queue, persistent configuration. |
| CAEVES Copilot Connector™ | Azure Container Instance | Indexes object content and ACLs into Microsoft Graph for Microsoft 365 Search & Copilot, Copilot Studio. |
| Identity & Access Management | Microsoft Entra ID / Active Directory | Entra ID, Active Directory, Kerberos, NTFS ACL preservation, managed identity issuance. |
Three deployment models
-
Public cloud. Instance and storage in the customer Azure tenant.
-
Hybrid. Instance on-premises or in colocation, object storage in Azure.
-
Edge to cloud. Distributed edge agents feeding a centralized Azure storage tier.
Metadata and data separation
Directory listing, browse, and search never touch the cloud. They are served from local metadata. Every file's bytes are an addressable object that lifecycle policies, redundancy settings, snapshots, and the connector operate on independently.
No proprietary backend
CAEVES writes standard Azure Blob objects with an open metadata catalog in Azure Tables. Read them with Azure Storage Explorer, AzCopy, or the Storage APIs, with or without CAEVES running.
INTELLIGENT DATA TIERING
Every file sits in the storage tier its activity justifies.
A policy engine built on Azure Lifecycle Management classifies file activity and moves content automatically. Every threshold is configurable at the share, folder, or policy group level, so retention behavior can follow project lifecycles and compliance obligations rather than one blanket rule.
Figure 2. Tiering and grounding operate independently. Cost falls as data ages, while search results are served from the Microsoft Graph index rather than the storage tier, so retrieval performance does not degrade with tier depth.
| Access tier | Default policy | Behavior |
|---|---|---|
| Hot | Modified within the last 30 days | Full performance for active working data. No disruption to end users. |
| Cool | Activity falls away | Lower storage cost with instant access. The landing tier for migrated data. |
| Cold | Not modified for one to six months | Further cost reduction, still instantly accessible. Per-GB retrieval fee applies. |
| Archive | Inactive for six months or beyond | Lowest-cost tier, approximately $0.002 per GB per month. Offline. Rehydration follows Azure Object Storage service levels. 180-day minimum retention applies. |
Files retain their original UNC path, NTFS permissions, and Windows Previous Versions restore points at every tier. Transparent recall returns content on access with no manual restore workflow.
Archive tier and search
Archive is an offline tier with no read availability SLA. In a CAEVES deployment, read availability is provided by the Copilot Connector through indexing and contextual search, so users can discover and evaluate archived content before a file is rehydrated. Metadata retrieval and search results are always instantaneous.
DATA PROTECTION AND RECOVERY
Backup, versioning, and immutability are part of the platform.
Snapshots and self-service restoreHourly VSS file snapshots give users Windows Previous Versions recovery with no IT ticket. Daily, weekly, and monthly full-volume snapshots run on a USN basis with tiered retention. Snapshots are sealed and read-only after creation. |
Control-plane recoveryMetadata snapshots to cold object storage, act as an independent recovery anchor. A full instance rebuild is possible without data loss because the metadata encodes the namespace and its binding to immutable objects. |
Ransomware postureWORM immutability can be applied to monthly and annual snapshot containers. Soft delete defaults to 30 days. Rapid rollback to a known-good snapshot avoids traditional archive restore workflows and supports aggressive RTO targets. |
Redundancy by recovery objectiveLRS for lowest cost and highest migration throughput. ZRS for in-region datacenter resilience. GRS and RA-GRS for regional-outage protection. Selected deliberately at provisioning, per workload. |
Get Started with CAEVES Intelligent Deep Storage™ in your own Microsoft Azure environment today.
FUNDAMENTALS - COPILOT CONNECTOR™
CAEVES Copilot Connector™
The same objects that make deep storage affordable also make it answerable. The CAEVES Copilot Connector™ indexes them into Microsoft Graph, in place, with permissions intact.
The CAEVES Copilot Connector™ deploys as an Azure Container Instance separately from the CAEVES instance. It reads the metadata process table as a distributed work queue, extracts and enriches content from object storage, embeds inherited ACLs, and registers CAEVES as an external data source in the Microsoft 365 admin center. No vector database. No embedding pipeline. No duplicate data silo.
01
Claim |
02
Extract |
03
Enrich |
04
Ground |
05
Serve |
| A claims-based processing model over Azure Table storage tracks pending, claimed, completed, failed, and expired jobs. The connector scales horizontally as volume grows. | Docling-based document parsing, OCR for scanned documents and images, Markdown normalization, and optional GPU-accelerated image, audio, and video indexing. | Structural metadata from ingest joins extracted content in the metadata process table. Custom fields support controlled vocabulary and classification taxonomies. | Content and inherited ACLs are ingested into Microsoft Graph as external items, in a partition owned by this connector alone. | Copilot, Copilot Studio agents, and Microsoft 365 Search retrieve from the index, cite the source document, and enforce permissions at query time. |
Key Benefits
Cross-document reasoning,
|
Scoped, staged, and governedPersistent filtering rules in the config table define which file types, source locations, and departments are indexed, and can be changed without redeploying. Staged rollout in the Microsoft 365 admin center releases the connector to pilot groups before full activation. Multiple agents can be deployed with distinct permission boundaries. |
Operational transparencyA health endpoint confirms environment configuration and Microsoft Graph SDK connectivity. A queue statistics endpoint reports pending, claimed, completed, failed, and expired indexing jobs in real time. A run-now endpoint triggers indexing on demand. Log Analytics captures the full audit trail. |
Legal integrity under AIAll AI processing occurs within the customer's own Azure tenant and Microsoft 365 environment, subject to the customer's own Microsoft data processing agreements. Responses are grounded in the unmodified archival record, protected by WORM immutability, snapshot sealing, and soft delete. |
Supported Content Types
| Indexed content | Formats |
|---|---|
| Supported Document types | PDF, DOC, DOCX, DOCM, PPT, PPTX, XLS, XLSX, XLSM, XLSB, RTF, ODT, ODP, ODS, TXT, Markdown, CSV, JSON, XML, HTML, MHT |
| Images | PNG, JPEG, GIF, TIFF, WEBP. OCR applied, with optional GPU-based image indexing |
| Email and messaging | EML, MSG, MHT, MHTML, NWS, Loop components |
| Archives | ZIP |
| Stored but not full-text indexed | Binary and proprietary engineering formats such as DWG, DXF, and SEG-Y. Stored, tiered, versioned, and retrievable. Not parsed for full-text search. |
WHAT'S NEXT
Ready to unlock the power of your data?
Get started with CAEVES in the Microsoft Marketplace in minutes. Migrate your first 5TB for free *.
* Limited Offer, consult the Microsoft Marketplace for terms & conditions