Table & Database

A table on a page holds structured rows for the people reading that page. A database is a separate store you can filter, view as a kanban board and reuse in more than one place. If the same records need to appear on several pages or drive an automation, build a database rather than a page table.

Relation fields connect a record in one table to records in another, and lookup fields pull a value across that link so nobody retypes it. Sub-tables let a single record carry its own set of rows, which fits cases such as a project with its tasks or an account with its invoices.

You can import into a table from a CSV file and export a table back to CSV, so a spreadsheet you already maintain can become a working database and a copy can go to whoever needs one. Duplicating a database or table is the quick route when a new project needs the same structure.

Yes. A database or dashboard can be embedded into a portal, so a client sees the records meant for them in their own space while the rest stays on your side. What each person can view or change follows the access you set, not the fact that the data lives in one system.

Agents can connect to FuseBase dashboards through MCP, a standard way of giving an AI tool access to a system, and can read and update records inside the permissions you grant. Give an agent the narrowest access that lets it do its job, and check what it writes before relying on it.

A general chat tool has no standing role, no access to your client records and no rules about what it may act on. An AI coworker in FuseBase is set up against real work, with scoped permissions, inside the process it is helping with.

You do, through the same permission system that governs people. An agent operates inside scoped access rather than above it, which is what makes delegating any of the client facing work reasonable in the first place.

Yes. Each client can have a portal of its own, branded to you, holding only the documents, updates and approvals that client should see. Access is set per person and per item rather than per workspace.

By putting the work somewhere it can be built on: structured records, clear permissions and a client facing surface that is already connected. Tools change quickly, but work that lives in a connected process is what a future tool can pick up.