Build a freelancer business dashboard: invoicing, payment tracking, and late-payment alerts

It's 11 PM on a Tuesday. You've finished a client project, sent the invoice three weeks ago, and you still have no payment. You fire up three different spreadsheets and an email folder, squinting at dates, trying to remember who owes you what. By the time you've chased down the payment, you've lost two hours you could have spent on actual work. This is the freelancer tax: time spent on admin instead of billable hours. The problem compounds. Each client uses a different payment method. One pays via bank transfer within thirty days. Another uses PayPal and forgets until you remind them. A third pays on NET60. Without a system, invoices slip through the cracks, payment cycles drag on, and your cash flow becomes unpredictable. You're running a business, but you're also running an unpaid collections department. The solution isn't a new invoicing tool. You probably have one already. The solution is connecting what you already use into a system that watches your invoices, tracks payment status, reminds you when payment is overdue, and alerts you before a client relationship becomes a collection problem. We're going to build that system with zero manual handoffs.

The Automated Workflow

We'll use n8n as the orchestration backbone because it runs locally and can handle complex conditional logic without vendor lock-in. Your invoice data will flow through AI-powered analysis, payment tracking, and automated reminders in a single, repeating workflow. Here's the architecture: invoices arrive in your email (via SaneBox), get analysed by Claude Sonnet 4.6 to extract payment terms and due dates, feed into a reminder system linked to your productivity app, and trigger payment alerts when invoices are overdue.

Step 1: Capture invoice data from email

Connect SaneBox to n8n using IMAP polling. SaneBox filters your inbox, so invoices land in a dedicated label. n8n checks that label every hour for new messages.

CopyGET https://mail.example.com/imap
SEARCH UNSEEN LABEL:"Invoices Sent"

For each new email, extract attachments and forward the message body plus metadata (sender, date sent, subject) to the next step. Store the raw email data in n8n's built-in database so you have an audit trail.

Step 2: Extract invoice details with Claude

Pass the email and any PDF attachments to Claude Sonnet 4.6 using the Claude API. The model is fast enough for real-time processing and strong at parsing semi-structured financial documents.

Copyjson
{ "model": "claude-sonnet-4.6", "max_tokens": 1024, "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Extract invoice number, amount, client name, invoice date, and payment due date from this invoice. Return JSON only." }, { "type": "document", "source": { "type": "base64", "media_type": "application/pdf", "data": "[base64-encoded PDF]" } } ] } ]
}

Claude will return structured JSON. Store this in a spreadsheet or database node so you have a master list of all outstanding invoices.

Step 3: Log payment status in your productivity app

Create a new reminder in your preferred productivity app (one with nested AI chats, like Notion or similar) for the payment due date. Set the reminder to trigger three days before the due date so you can send a friendly reminder to the client.

Copy[POST](/tools/post) /api/reminders
{ "title": "Payment due: Invoice INV-2026-001", "client": "Acme Corp", "amount": "£2,500", "due_date": "2026-04-15", "alert_date": "2026-04-12", "status": "pending"
}

Step 4: Run a daily payment status check

Each morning at 9 AM, trigger a second workflow that: 1. Fetches all pending invoices from your database 2. Calculates days overdue for each invoice 3. Checks your bank account balance (via Payman AI or a direct banking API if available) to flag cash flow concerns 4. Uses Claude Haiku 4.5 (fast and cheap) to draft follow-up emails for overdue invoices

Copyjson
{ "model": "claude-haiku-4.5", "max_tokens": 512, "messages": [ { "role": "user", "content": "Client: Acme Corp. Invoice: INV-2026-001. Amount: £2,500. Due date: 2026-04-15. Days overdue: 7. Draft a professional payment reminder email. Keep it to 100 words." } ]
}

Step 5: Alert on payment received

When a payment lands in your bank account, trigger a webhook (or poll your bank's API every few hours). Match the payment amount and date to your invoice list. Mark the invoice as paid, close the reminder in your productivity app, and remove it from your overdue report.

Step 6: Weekly dashboard summary

Every Monday morning, generate a summary report showing: - Total outstanding revenue

• Invoices overdue by 30+ days
• Average days to payment
• Clients with the worst payment history Use Claude Opus 4.6 to generate a brief narrative analysis. This becomes your weekly financial health check without you having to open a spreadsheet.

Copyjson
{ "model": "claude-opus-4.6", "max_tokens": 800, "messages": [ { "role": "user", "content": "Analyse this invoice data and identify cash flow risks: [JSON data]. Highlight the top three concerns and suggest actions." } ]
}

The Manual Alternative

If you prefer more control, you can run this workflow less frequently or stop at the reminder stage. Use a spreadsheet with manual formulas to track invoice status, then use SaneBox filters to surface overdue invoices in a dedicated folder you check weekly. This removes automation but keeps the organisation. Alternatively, if your invoicing tool has a native integration with your email and banking system (many modern tools do), you can skip n8n entirely and rely on built-in rules. The trade-off is less flexibility and higher per-tool costs.

Pro Tips

Avoid duplicate reminders.

When Claude extracts invoice details, check against your existing database before creating a new reminder.

Use a unique invoice number as the key to prevent duplicate entries.

Set conservative rate limits.

Claude Sonnet 4.6 and Haiku 4.5 are fast, but if you're processing 50+ invoices daily, stagger requests across 5-minute windows to avoid hitting API limits. n8n's rate limiter node handles this automatically.

Monitor for payment fraud.

If a payment matches an invoice amount but arrives from an unexpected account, flag it for manual review. Claude can be trained to detect suspicious patterns; add a simple check: "Does this payment source match the client's usual method?"

Archive old invoices.

After 12 months, move paid invoices to cold storage. Keep active invoices (unpaid or recently paid) in your hot database so queries run fast.

Test with a single client first.

Wire up the workflow for one trusted client, run it for a week, then expand. This catches configuration errors before they affect your entire client list.

Cost Breakdown