Varen Documentation
This is the reference for using Varen. If you're new, start with Getting Started. For background on what Varen is and how it's designed, read on. For marketing context and the full product overview, visit varen.au.
What is Varen?
Varen is a hybrid AI research and analysis platform. It gives you persistent workspaces, conversational assistance, evidence-backed Lens research, local knowledge and specialist routing across domains including finance, tax, health, property, law, business and more.
The defining characteristic of Varen is the separation between the node data plane and the SaaS control plane. Conversations, uploaded files, knowledge and local indexes are durable node content. SaaS coordinates authentication, node connectivity, routing, usage and billing; it is not the customer content store. This does not mean every request is local: hosted inference can receive the question and relevant retrieved context.
The practical result is a platform that accumulates useful workspace context on your node and applies specialist knowledge to it, with an explicit trade-off when a hosted model is selected.
Three levels of organisation
Varen organises everything into three levels: nodes, workspaces, and sessions. Understanding why the hierarchy exists — not just what each level is — makes the platform much easier to use effectively.
A node is a hardware instance running the Varen software. It could be a machine you own at home, a cloud host you control, or — where available — a managed instance. The node is where your durable data lives: all workspaces and sessions on a node are stored on that hardware.
A workspace is an isolated project or topic area. The isolation is intentional. Varen builds up a detailed picture of each workspace over time — the facts you've stated, the documents you've uploaded, the knowledge it has inferred. If all of that accumulated in one place, different areas of your life would contaminate each other. Your tax research would colour your health conversations. Your work planning would bleed into family decisions. Workspaces isolate context by default; explicit linked-workspace features can allow selected workspaces to contribute retrieval context.
A session is one conversation thread within a workspace. Sessions can be resumed at any point. Multiple sessions in the same workspace share the same knowledge base and files, but the conversation threads remain separate.
| Level | What it is | Scope of memory |
|---|---|---|
| Node | One hardware instance running the Varen software | All data on the node is on this hardware |
| Workspace | An isolated project or topic | Facts and files are private by default; explicit links can extend retrieval scope |
| Session | One conversation thread | Stored within its workspace; can be resumed |
How to use these docs
If you're new to Varen, Getting Started walks you through your first login, adding a node, and starting your first conversation. After that, the guides cover each part of the platform in depth.
For existing users, use the sidebar to go directly to the relevant guide. Workspaces & Sessions covers how to organise your work. Chat & Advisers covers the conversation interface and how routing works. Knowledge & Profile explains how Varen's memory works. Files & Resources covers uploads and downloads. The Reference section covers adviser domains, billing, and privacy.
A note on privacy
Your durable content — conversations, files, knowledge base — is stored on your node. To generate a reply, the control plane assembles the prompt and calls a model provider, so your question and the relevant retrieved context cross the SaaS and provider boundaries for that turn. See Privacy & Data for the full picture.