In most finance teams, the week starts the same way. Someone downloads the bank statements, saves them in a folder and matches payments against invoices by hand. The documents come from everywhere — a shared mailbox, the bank, the ERP, a file server — and each of those systems has its own login.
Usually these systems get connected one script at a time. Passwords end up in configuration files, nobody knows which process depends on which account, and when a bank token expires, a nightly job quietly stops.
Our new video guide shows how DocReveal handles this with the Connection Hub. It runs under seven minutes, you can watch it at the top of this page, and it follows Keldmoor, a fictional distributor, from the first connection to answers in chat. Here is what you will see and how it works.
One place for every connection
The Connection Hub is where DocReveal connects to your other tools. It opens from the plug icon in the header and has four tabs: your connections, the service catalog, custom connectors and activity.
Every connection shows its state, how long it is valid and who uses it. That changes how you work: before you touch a connection, you know which flows and agents depend on it. And while an enabled flow uses a connection, DocReveal does not let you delete it.
Connecting a service takes one dialog
The service catalog lists apps, banks and custom connectors, and you can filter it by type, sign-in method or category. Pick a service and DocReveal tells you how you sign in, what you will need and what the service can do. A mailbox over IMAP, for example, comes with four actions and a trigger for new e-mail.
The connect dialog then asks for exactly that: a name, the server and the sign-in. Secrets are stored encrypted and never shown again. You connect a service once, and your flows and agents use that connection from then on.
Your bank, read-only
Banks have their own filter in the catalog, organised by country. The connection uses access from your internet banking: you paste the API token issued there, together with the date it is valid until. DocReveal then loads your accounts — in the video, an operating account in dollars and an account in euros.
The bank connection only reads. DocReveal sees accounts, balances, transactions and statements, and it cannot make a payment. Before the token expires, DocReveal reminds you.
Practical use: the bank statement that files itself
This is the heart of the video. In Integration Manager, a flow called Daily bank statement runs on a schedule, every day at 7:00 UTC, in four steps:
- The schedule starts the run each morning.
- A Date Helper step works out yesterday’s date.
- The bank step, Statement for a period, downloads that day’s statement for the operating account.
- Upload document to DocReveal files it in a Bank folder, one subfolder per month, with the label Bank statement.
From there, each statement is processed like any other document. DocReveal recognises it as a bank statement, writes a short summary with the day, the movements and the closing balance, turns every transaction into data and adds it to the knowledge graph. Every run is logged step by step, down to the file it uploaded.
Practical use: which invoices did we actually pay?
Because the statements now sit next to the invoices, the team can ask about both at once. In RAG chat, it asks which supplier invoices were paid last week and which are still open.
The answer matches statement lines to invoices: one invoice paid as approved, a Czech supplier paid in euros, and one payment for which DocReveal holds no invoice at all. It lists what is still open, including three disputed invoices on hold, links every point to its source and suggests the next step: ask that supplier for the missing invoice.
These are the same September invoices you may know from Ask your invoices and Agents under control. This time, the bank statements join them on their own.
Practical use: agents that act through your connections
Agents use the same connections. When you give an agent its tools, you can pick actions from connected services and from your own connectors. In the video, the Invoice triage agent can send e-mail through the AP outbox and look up a purchase order in the ERP.
Sending e-mail changes data, so the tool is marked sensitive and follows its group’s policy — here, a person has to approve it first. A mailbox can also start an agent: the Mailroom agent checks the AP mailbox every five minutes.
Practical use: connect what isn’t in the catalog
When the system you need has an HTTPS API but isn’t in the catalog, you describe it once in Connector Studio. In the video, the ERP connector is imported from the ERP’s OpenAPI file. You set the service address — the connector may only call that host — and the sign-in, here an API key in a header.
Each action is one call to the service. You test it with the real connection, and only a successful test unlocks publishing. Once published, the actions appear in the flow designer and in agent tools: Keldmoor’s approved invoices now become payables in the ERP through the new connector.
Connections that stay healthy
Bank access does not last forever. Before a token expires, the connection moves to Needs attention and Integration Manager shows a reminder. Renewing asks for the new token, and the connection and every flow using it stay in place. The activity log records who created, renewed or published what, and when — including the system’s own expiry warnings.
What to take away
- Connect once. One connection serves every flow and agent that needs it, and you always see who uses what.
- Keep the bank read-only. DocReveal reads statements; payments stay where they are today.
- Let documents arrive on their own. A scheduled flow turns yesterday’s statement into a searchable document every morning.
- Ask, then act. Chat matches payments to invoices with sources, and sensitive agent actions can wait for a person’s approval.
Keldmoor, its banks and its ERP are fictional demo data; the screens are the real product.
Connect once. Use everywhere.