Solution
Private AI knowledge assistant.
A private, self-contained AI assistant for the knowledge you choose to share with it. Spin it up in your own infrastructure — cloud, on-prem, or ours — give it a defined set of documents, and get fast, cited answers without your data ever leaving your environment.
- Private to your environment
- Cited answers, not guesses
- SSO + audit log built in
What it does
Three things, well.
Answer staff questions instantly
Plug in your policies, runbooks, contracts, regulations, or product manuals. Staff get accurate answers in seconds — not "I'll get back to you tomorrow".
Cite the source document
Every answer points to the document and page it came from. People can verify what they're reading. Auditors can trace the trail.
Never leak to a public LLM
The assistant runs in your cloud, your data centre, or on infrastructure we manage on your behalf. Your documents never train a third party's model.
How it works
Own your AI.
Three steps. Your data stays yours.
-
Point it at your documents.
PDFs, Word files, SharePoint, intranet pages — whatever your team already relies on. We curate the source set with you so the assistant only sees what it should.
-
It learns inside your environment.
Your documents are indexed in a private database that runs in your environment — your cloud, your on-prem, or hosted by us. Nothing is sent to a third-party model provider unless you choose to.
-
It answers — with citations.
Staff ask questions in plain English. Every answer shows the source document and page number, so the reader can verify it in one click. Wrong answers become improvable, not invisible.
Built on retrieval-augmented generation (RAG). Runs entirely inside your network when required.
Capabilities
Everything an enterprise expects. Nothing a buyer doesn't.
Knowledge Assistant ships with the controls regulated and security-conscious organisations require — identity, audit, content safety, and the ability to run entirely inside your own network.
Differentiator
Data Curation & Model Tuning
A productised assistant is only as good as the knowledge it's given. As part of every deployment, our team reviews your source documents for accuracy, currency, and contradictions before ingestion; sanitises sensitive content (strips or redacts PII, commercially sensitive figures, out-of-scope material); tunes retrieval (chunking strategy, embedding model selection, re-ranking for your document types and query patterns); optionally fine-tunes the language model on your domain (regulatory framework, internal terminology, house style); and establishes an evaluation harness — a test set of representative questions with expected answers — so you can measure quality before and after every change.
This is the difference between a chatbot and an assistant your staff actually trust.
What's in the box
-
Knowledge you control
Drag-and-drop ingestion of your PDFs, Word, Excel, SharePoint and intranet content. Admins see exactly what's indexed and can remove anything, any time.
-
Choice of intelligence
Run a fully self-hosted model, an Azure / AWS / GCP enterprise model, or a frontier API. Mix and match per workspace.
-
Enterprise sign-on
OIDC and SAML out of the box — Entra ID, Okta, Google Workspace, Ping, ADFS. Map your existing security groups straight to roles.
-
Multi-workspace by design
One deployment, many isolated assistants — one per team, department, or knowledge domain. No cross-contamination, separate access lists.
-
Cited answers, audited interactions
Every answer cites the source document and page. Every interaction is logged for compliance, Subject Access Requests, and your SIEM.
-
Safe by default, deployable air-gapped
Layered content safety, prompt-injection and PII filtering, no outbound calls unless you configure them. Runs offline if your environment requires it.
Deployment models
Three ways to run it.
Same software. Choose where it lives and who operates it.
- 01
In your cloud
Volorai installs into your Azure, AWS, or GCP subscription using Terraform modules we provide. You pay compute. Volorai charges a fixed subscription per platform instance. Your data never leaves your environment.
- 02
In your data centre
Kubernetes (AKS, EKS, GKE, OpenShift, vanilla K3s) or Docker Compose on a single GPU host. Air-gapped deployments supported. Suitable for regulated, sovereign, or sensitive environments.
- 03
Managed by Volorai
We run it on UK-sovereign infrastructure on your behalf. Single invoice, faster onboarding. Your documents are processed in a dedicated, isolated environment — ask us for the data processing agreement.
Tech Stack
Built on proven open-source foundations.
No proprietary lock-in. Every component is replaceable, auditable, and deployable in your environment.
Web tier
- NGINX
- React
APIs
- Node.js
- Express
- FastAPI
Inference
- Llama.cpp
- vLLM
- TurboQuant
Queue
- BullMQ
- RabbitMQ
Stores
- PostgreSQL
- MariaDB / MySQL
- LanceDB
- Chroma
Runtime
- Docker
- Kubernetes
SCM
- GitHub Enterprise
OS
- Ubuntu
- CentOS
- RHEL
Proof
In use today.
Knowledge Assistant has been deployed for a UK professional body and is in preparation for a regulated organisation deployment.
Frequently asked questions
-
Does my data leave my organisation?
On the customer-hosted deployment model, no — your data stays in your cloud or on-premises infrastructure throughout. On the Volorai-hosted model, data is processed in a dedicated, isolated environment; ask us for the data processing agreement.
-
How long does deployment take?
A standard customer-hosted deployment to an existing Azure, AWS, or GCP subscription typically takes 1–2 weeks from access grant to live assistants. On-premises or complex network configurations take longer.
-
Can we configure multiple assistants for different teams?
Yes. The platform is designed for multi-assistant operation. Each assistant has its own knowledge corpus, access controls, and system configuration. You can have separate assistants for legal, finance, IT, and client support — all on the same platform instance.
-
What AI models does it use?
The platform supports any OpenAI-compatible API — including Azure OpenAI, Anthropic, and self-hosted models. Model selection is configurable per assistant. For air-gapped or sovereign deployments, we can configure fully local inference.
-
What does the subscription include?
Volorai installation, configuration, and ongoing support. Platform updates. Access to Volorai for configuration changes, troubleshooting, and new assistant onboarding. Compute is separate (yours on customer-hosted; included on Volorai-hosted).
-
How does this compare to Microsoft Copilot or similar SaaS options?
Microsoft Copilot and similar SaaS AI tools send your data to a cloud provider's shared infrastructure and are priced per user. Knowledge Assistant is private to your environment, priced per platform (not per user), and configured specifically for your documents and access patterns — not a general-purpose assistant. The cost comparison is significant at scale.
Bring a use case. We'll show you the deployment.
A 30-minute demo using your own example documents. We bring a working environment; you bring a folder of representative content. By the end you'll know whether Knowledge Assistant fits.