Skip to content

Data Architecture

Each workspace is a separate data environment. Switching between them is not a change of view. It changes the entire set of data the system is operating on.

The boundary is the workspace

A workspace holds everything the firm has put into HAQQ: matters, contacts, documents, financial records, timesheets, AI context and settings. That set is the unit of isolation. The dashboard, matters and documents you see in one workspace do not appear in another, and that is by design rather than a synchronisation delay. Each workspace is its own data silo, which is what makes complete separation between offices, entities and jurisdictions possible.

Knowledge never crosses a workspace

Knowledge is isolated by workspace at the strictest level in the product. Documents uploaded in one workspace are never reachable from another, even by the same user holding membership of both. Inside a workspace there are two pools, and only one of them is shared.

The two knowledge pools
PoolWhat is in itWho can reach it
Company knowledgeFirm-wide documents, all matters, contacts, calendar events, tasks and billing records.Every workspace member whose role permits it.
Personal knowledgeYour own AI conversations, personal files and private matter notes.Only you. Never shared with other workspace members.

Your knowledge base is not used to train AI models. The AI reads it at query time and nothing is retained outside the workspace boundary. Legal libraries covers how a firm builds and curates that pool.

In HAQQ Legal AI Chat, personal details in a conversation are replaced with made-up stand-ins before anything is sent to an external AI model provider. Pseudonymisation shows the six steps and what the model provider receives.

What is shared inside one workspace, and what is not

  • Every seat gets its own AI and its own private history. You cannot open a colleague's conversations and they cannot open yours.
  • Credits work the other way. They sit in one workspace pool that everybody draws from, so a firm with several seats has one shared balance and one heavy user can draw most of it down. Worth knowing when you size a plan, and worth saying out loud to the team.
  • Usage quotas and plan limits are enforced at workspace level, not per person, for the same reason.

Credits and usage explains the pool and its top-ups in full.

One workspaceCompany knowledgeEvery member whose role permits itMatters, contacts, calendar, billingSeat 1Own AI historyOnly them, never sharedSeat 2Own AI historyOnly them, never sharedSHARED CREDIT POOLOne balance, every seat draws from itAnotherworkspaceKnowledge never crosses a workspace, even for a member of both
One workspace boundary holding shared company knowledge, private per-seat personal knowledge and one shared credit pool, and nothing crosses into another workspace.

Access inside the boundary

Isolation separates workspaces from each other. Inside one workspace, who reaches what is a role question, on three levels: which modules a role can act in, which records inside a module it reaches, and which fields it sees on those records. Monetary amounts on the chart of accounts, for example, can be hidden from a role that still has to file against it. Setup Roles & Permissions is where all three are configured. Removing a member revokes their ability to act in the workspace while leaving their name on the work they did.

Encryption, residency and compliance

Platform-level guarantees (encryption at rest and in transit, audit logging, data residency, retention policy and HAQQ's compliance position) are published and kept current on the HAQQ security page. They are stated there rather than duplicated here, so there is one version of them to keep accurate.

Tópicos relacionados

Esta página foi útil?