Introducing AI Autopilot

Today, we’re releasing our biggest feature yet: AI Autopilot.

Since starting Briefcase in 2024, our ambition has been to deliver end-to-end automation in the core bookkeeping and accounting workflows that frequently drain time and talent: from invoice capture to categorisation, VAT compliance and financial close.

The need for this automation has never been greater. Client expectations are rising, margins are tight and firms are struggling to scale profitably.

With our agentic platform, we’ve already delivered meaningful improvements, cutting invoice processing tasks down from average 60 seconds to 20 seconds. But the “last mile” has always required manual intervention.

Until now.

What is AI Autopilot?

AI Autopilot reviews every transaction uploaded into Briefcase and decides whether it can be published straight to the ledger or whether it needs your review. This means a large proportion of invoices and receipts can be autopublished, with no manual work required.

Autopilot runs after all AI agents have finished their processing. It then applies a set of deterministic checks to decide which transactions are safe to publish. There’s no need for manual training, configuration or supplier rules — and every decision can be easily audited in the product.

How does it work?

Autopilot runs three verification checks before making a decision:

1. Document legibility

2. Document completeness

3. Historical consistency

If any one of these checks fails, the transaction won’t be autopublished.

If all three pass, it will be autopublished.

1. Document legibility

The first test checks that the document is actually readable. Invoices and receipts can vary widely in quality — from sharp PDFs to faded scans and handwritten notes. If the content isn’t clear or leaves room for ambiguity, it’s not safe to rely on.

To avoid mistakes, Briefcase extracts the data twice. If the results don’t match, the transaction won’t be autopublished.

2. Document completeness

The second test checks that all key data fields, such as the date, due date, reference, amount, account code and tax rate, are present.

As part of this test, we also implement the following guardrails:

• All line items must add up to invoice total.

• Supplier must be existing supplier.

• Document type must be invoice, receipt or credit note.

If a required field is missing, or one of these guardrails fails, the transaction won’t be autopublished.

3. Historical consistency

The final check confirms that the transaction is historically consistent with what you and your team have published in the past. AI Autopilot compares the document against historical transactions, checking that key fields such as the supplier, category, VAT rate, tracking codes and publish location are consistent.

We require at least three consistent historical transactions before autopublishing. Older or edited transactions are weighted less, so the system prioritises your most recent and up-to-date behaviour.

This approach means Autopilot continually adapts to your workflow over time. If you edit an autopublished transaction, similar ones will be less likely to be autopublished in future. If you consistently approve them without changes, Autopilot grows more confident publishing those automatically.

How can I try it?

AI Autopilot is live today for all Briefcase users. Get in touch if you want to give it a go.