Hardware options
Rent, buy, or bring your own — deployment maps to your preferred ownership model.
PrivateBox
platform.
A private AI operating layer for confidential work.
Deploy a full in-house AI environment on hardware you control or hardware we help provision. PrivateBox combines private inference, secure enterprise search, governed access, connectors, and custom workflows inside one ring-fenced platform.
Private inference
Secure enterprise search
Governed access
Flexible deployment
Ring-fenced environment
One platform.
flexible deployment.
PrivateBox is built for organisations that need more than a chatbot and less than a patchwork of vendors. It provides a controlled AI environment for confidential work, internal knowledge access, custom workflows, governed usage, and long-term flexibility.
A full private AI stack — not a single feature wrapped in marketing.
Platform software
A shared core platform that stays consistent whether you run one node or many.
Private inference
Run AI inside a controlled environment instead of relying on public infrastructure.
Governed access
Role-based controls decide who can use tools, data, workflows, and actions.
Secure enterprise search
Search internal documents with cited answers and controlled visibility.
Workflow flexibility
Custom actions, connectors, and bespoke tooling on the same foundation.
Architecture.
shown clearly.
A layered private AI environment built for controlled deployment, governed use, and long-term flexibility. Each layer plays a distinct role — hardware at the base, workflows on top.
Not a single feature. A full stack designed for confidential work.
Layer 01
Hardware
PrivateBox hardware
Client-owned hardware
Modular node architecture
Internal networking
Layer 02
Operating
Controlled access
Private inference runtime
Management layer
Secure connectivity
Internal operations
Layer 03
Platform
Private chat
Secure enterprise search
Role-based access
Connectors
Agent creator
Custom actions
Governed workflows
Layer 04
Workflow
Internal knowledge
Drafting and review
Email workflows
Meeting intelligence
Bespoke tools
Custom AI apps
Flexible deployment.
built to fit.
PrivateBox can be deployed in different ways depending on how your organisation prefers to adopt private AI infrastructure.
Same core platform. Same flexible architecture. Different ownership paths.
Rent hardware
Start with PrivateBox hardware in a lower-friction deployment model while keeping the same core platform and support path.
Best for lower upfront commitmentBuy hardware
Own the hardware from day one and deploy the PrivateBox platform inside infrastructure allocated to your organisation.
Best for long term internal ownershipBring your own hardware
Deploy the PrivateBox platform on approved client-owned infrastructure while maintaining the same product foundation and flexible architecture.
Best for teams with existing infrastructureFor larger environments, PrivateBox also supports more tailored in-house deployment paths and modular expansion over time.
Core capabilities.
built in.
A full platform of capabilities designed for confidential work. Tap any card to reveal a more technical view.
Twelve capabilities. One controlled environment.
Private Chat
Confidential day-to-day AI work inside your own environment.
Session boundaries stay inside your deployment; logs, context, and retention follow your policy — not a vendor's.
Secure Enterprise Search
Search internal documents with cited answers and controlled access.
Indexed approved content, cited retrieval, role-bound visibility — answers always trace back to their source.
Agent Creator
Build governed internal agents around real business roles and workflows.
Define data sources, permitted tools, and review paths; agents run inside a predictable, auditable boundary.
App Connectors
Connect selected internal systems and data sources into the platform.
Pre-approved integrations surface context where teams already work — without collapsing the security boundary.
Private Inference
Run AI inference inside a controlled environment instead of relying on public infrastructure.
Inference runs inside your deployment so prompts, context, and outputs never leave your environment.
Role Based Access
Control who can access tools, workflows, knowledge, and actions.
Define roles once, apply everywhere — from document visibility to action execution and agent scope.
Custom Actions
Move from passive prompting into structured operational workflows.
Actions encode business operations the way internal teams actually run them — not as freeform prompts.
Custom AI Apps
Deploy focused tools for specific internal use cases.
Small, focused apps for specific jobs — not one generic interface trying to cover every workflow.
Governed Workflows
Use AI with structured internal controls instead of ad hoc usage.
Controls travel with the work: approvals, audit trails, and policy bindings live inside each workflow.
Open Model Flexibility
Maintain freedom to evolve the stack without unnecessary vendor lock-in.
Your platform evolves with the model landscape — not stuck behind one provider's release cadence.
Optimised Inference
Use a tuned stack built for practical performance and maintainability.
The stack is tuned for real-world throughput, latency, and long-term maintainability — not demo numbers.
Ring Fenced Environment
Keep AI activity isolated within a controlled organisational boundary.
One controlled perimeter — inference, data flows, and activity stay inside the organisation using it.
Same platform.
different scale.
Every PrivateBox deployment starts from the same core platform foundation. The difference is how it is deployed, how it is supported, and how it is scaled inside the organisation.
Hardware determines shape and expansion. Platform tiers mainly reflect support depth and environment needs.
Deployment shape and expansion
- Rent
- Buy
- Bring your own
- Modular node scaling
Support depth and operating model
- Shared core platform
- Support tiers
- Maintenance
- Future-proofing
- Customisation where needed
How teams.
use it.
Abstract platform descriptions only go so far. In day-to-day use, PrivateBox shows up as a handful of very concrete patterns.
Real operational usage. Not demo theatre.
Business Brain
Turn internal knowledge into an assistant that understands your company context, processes, documents, and approved information.
Draft and Review
Use PrivateBox for confidential drafting, revision, and internal review workflows — without routing material through public tools.
Email Triage
Create internal agents that filter, summarise, and structure important communication so the right things surface first.
Meeting Intelligence
Capture and summarise meetings inside a private AI operating environment — no third-party transcription middleman.
Focused Tools
Deploy focused AI tools for specific internal tasks instead of relying on one generic interface to handle everything.
Built for.
confidential industries.
PrivateBox is relevant wherever confidentiality matters. The same platform, applied with the shape of each field.
Pick a vertical to see example internal workflows.
Adoption support.
when needed.
PrivateBox also supports training, workflow design, bespoke tools, custom integrations, and phased rollout planning for organisations that need more than deployment alone.
Hands-on help where it matters — from onboarding teams to shaping bespoke internal workflows. Product-first, with the support layer right behind it.
Training
Adoption strategy
Custom workflows
Bespoke tooling
Governance.
built in.
PrivateBox is designed for ring-fenced deployment, controlled access, encryption in transit, and compliance-oriented operating practices.
For deeper architecture, controls, and roadmap detail, view the dedicated Security overview.
Ring-fenced deployment. Controlled access. Encryption in transit. Governance-aware by design — with a dedicated Security page for anyone who needs to go deeper.
View Security Overview
See PrivateBox
in action.
Explore the platform in a focused product demo built around your workflow, your internal environment, and your confidentiality requirements.