Onboarding and Expense Automation
Mercury’s onboarding streamlines the KYC process using AI. You must provide your legal business entity information, EIN, and business address. The system cross-references this data with federal databases in real-time. You must have your formation documents, such as your Articles of Incorporation or LLC Agreement, ready in PDF format. The AI parser extracts relevant details from these documents to reduce manual entry errors. I found the process takes about 5 to 10 minutes once you verify the extracted data. Mercury’s manual review process causes annoying delays if your business name does not match your legal documents exactly. Once you log in, head to the Automation tab to set up rules for transaction categorization and alerts. You can define categories for primary expense types like ‘Cloud Infrastructure’, ‘Software Subscriptions’, or ‘Payroll’. I would start with broad categories and refine them after processing your first month of transaction data. The AI engine learns from your initial inputs. You can also create conditional rules, such as telling the system to flag any transaction over $500 to AWS as Critical. The system uses merchant identifiers to auto-categorize future transactions. For advanced users, Anomaly Detection uses machine learning to identify unusual spending patterns that indicate fraud or errors.
API Integration and Security Controls
Mercury provides a REST API for custom setups. You can write a simple webhook listener to push transaction data to your internal database to eliminate manual reconciliation, but you must ensure you map Mercury’s transaction types to your accounting software’s chart of accounts correctly to avoid reporting errors. Developers use the GET /accounts endpoint to retrieve account balances in their applications. You can subscribe to events like transaction.created or balance.updated to trigger custom actions in your backend. To avoid hitting API quotas during high-traffic periods, you should implement rate limiting in your client code.
| Feature | Action/Endpoint | Purpose |
|---|---|---|
| Accounting | Settings > Integrations | Connect QuickBooks Online or Xero via OAuth |
| Security | Security > Approval Workflows | Define threshold limits for signatures |
| API | GET /accounts | Retrieve account balances |
Mercury provides native integrations with platforms like QuickBooks Online and Xero. Go to Settings > Integrations to select your preferred platform and authorize the connection via OAuth. This bi-directional sync ensures every transaction in Mercury appears in your general ledger. To manage security, go to Security > Approval Workflows. You can define threshold limits, such as requiring two signatures for any payment over $1,000. Mercury assigns roles like Admin, Approver, or Viewer to team members. The AI suggests approvers based on historical patterns and role responsibilities. When a transaction starts, the AI checks for anomalies and routes it to the appropriate approvers. Notifications reach users via email and in-app alerts. The system automatically pulls funds once the approver approves the workflow. I recommend you enable 2FA for all users with approval privileges to increase security.
Forecasting and Accounting Sync
Manual reconciliation wastes time, as you already know. Go to Settings > Integrations to connect Mercury to QuickBooks Online or Xero via OAuth. This bi-directional sync ensures every transaction in Mercury appears in your general ledger. You should test the connection by making a $0.01 transaction and checking if it appears in your accounting software. The Insights dashboard provides AI-driven cash flow forecasts for the next 30, 60, and 90 days. The AI model analyzes your historical transaction data, recurring expenses, and expected inbound payments. It accounts for seasonality and recent spending trends to provide a dynamic view of your financial health. You can adjust variables manually, such as adding a large expected invoice, to see how it impacts your forecast. The forecast updates automatically as new transactions occur. You can use ‘What-If’ scenarios to test different financial strategies. How does the AI handle a sudden, massive surge in unexpected subscription costs?
