Connections

Everything in sanda starts as rows in your warehouse. There are three ways to put them there.

Connections are the main way in, and most of this section is about them.

What a connection is

A connection is a saved link between one source and your warehouse. It records four things:

• The source, and the details sanda needs to sign in to it.
• What to sync: which tables the source offers, called streams, and how each one is read (its sync mode).
• When to sync: a schedule, or manual.
• Where it lands: which warehouse the rows go to.

sanda's managed sync does the reading. The details you enter go straight to it over TLS, and sanda's own database never stores them. You never paste a warehouse connection string: sanda wires the destination itself.

Where rows land

Your warehouse is a standard PostgreSQL database. Rows from a connection land in its raw schema, one table per stream, named raw.<connection>__<stream>. <connection> is the connection's name in lowercase, with each run of other characters turned into one underscore and the result cut at 40 characters. A connection named Acme · Xero (AU) lands its invoices in raw.acme_xero_au__invoices.

Two connections whose names reduce to the same prefix write to the same tables, so keep names distinct once the punctuation is gone.

Landing does not put anything on your semantic map. Once rows have landed, ask cherry to learn from your data, or add the tables yourself. See Learn from your data and Explorer.

The connections list

Open Data · Connections to see every connection.

Each row shows the connection's name and source, where it lands, its frequency, when it last synced, the rows it has synced in total and its status. A row that is syncing shows the run's progress in place of the total. The search box filters by name or source.

Three buttons sit above the list:

• New connection starts the add a connection flow.
• Load a CSV opens the CSV upload.
• Sync times opens the account schedule, which draws every scheduled sync, task and delivery on one clock. See Sync schedules.

Under the list, sanda reminds you of the fixed address it syncs from. See Allowlist sanda's address.

Connection statuses

Every connection shows one status, on the list and at the top of its own page.

Connectors offered by arrangement

The catalogue lists the systems sanda can connect to. Most have a setup form you fill in yourself. A few are set up with you instead: they carry a Concierge badge, and there is nothing to fill in beyond a name and a frequency. A sanda engineer wires the source and confirms the first load with you.

• MYOB
• Something else

Something else is for a system that is not in the catalogue. Create the connection, describe what you need, and sanda takes it from there. See the full connector catalogue for every source sanda offers.

Rows an assistant adds

An owner or admin can open a CSV table for intake. A connected assistant holding the intake permission can then add rows to that table, matching its columns, and can do nothing else to it. It is meant for a log someone keeps by talking: a daily log, a form, a running list. You open a table for intake when you load the CSV, and close it in the Intake tables panel on the Settings page. See Agent tokens and MCP tools for the assistant's side.

Keep reading