For finance teams adopting AI

The trust layer for AI work.

Relay turns AI-prepared finance actions into approval-ready Capsules — with evidence, permissions, risk levels, human approval, and audit logs built in.

The problem

AI can prepare the work. But companies still need control.

AI can now draft invoice follow-ups, refund decisions, supplier replies, and customer messages. But serious businesses cannot let AI act freely without knowing what it was allowed to do, what evidence it used, who approved it, and whether it can be audited.

The bottleneck is no longer AI capability. The bottleneck is trust, control, and accountability.

The solution

Every AI action becomes a Capsule.

A Capsule is a structured record of an AI-prepared action. It gives finance teams a clear approval object before anything sensitive is sent, changed, or executed.

Proposed action

Supporting evidence

Risk level

Required permissions

Human approval

Audit trail

First use case

Invoice exception follow-up.

Relay starts with messy unpaid invoice cases where context is spread across spreadsheets, emails, payment notes, and internal decisions.

Example capsule

INV-882

Customer: Supplier AB

Amount: 42,000 SEK

Issue: Customer claims the invoice is missing a purchase order number

AI recommendation: Send a corrected invoice copy and request confirmation

Risk level: Medium

Awaiting human approval

Workflow

How Relay works

1

Upload invoice data

2

Add customer context

3

Relay prepares the action

4

A human approves

5

Everything is logged

Not another invoice reminder tool.

Relay is building the control layer for AI work inside business-critical workflows. Today: invoice exception Capsules. Next: supplier approvals, refund approvals, procurement actions, and contract follow-ups.

Want to safely test AI in your finance workflow?

We are currently speaking with finance teams, accountants, and small business operators handling unpaid invoices, disputes, and customer follow-ups.