Invoice

Invoices are generated informally or not at all. The founder sends customers a PayPal request or a PDF created in Google Docs with no numbering scheme. Some customers pay based on verbal agreements without ever receiving a formal invoice. When the auditor asks 'show me the invoice for this revenue' there is nothing to show.

None — AI cannot perform any invoice analysis, revenue tracking, or collections optimization because no invoice records exist in any system.

Invoices are generated from the billing tool with basic fields — customer name, total amount, and due date. But line-item detail is inconsistent: some invoices show 'Platform Subscription $5,000' as a single line, others break out base fee and per-seat charges. Tax calculations are manual. Matching an invoice to the underlying subscription contract requires opening both records side by side. When a customer disputes a charge, the finance team spends hours reconstructing what was billed and why.

AI could flag overdue invoices and compute basic aging totals, but cannot perform automated subscription-to-invoice reconciliation because invoice line items do not consistently map to subscription components.

Invoices are generated from the billing platform with standardized line items — base subscription, seat charges, add-on fees, proration credits, and calculated tax. Each invoice references a subscription ID. Finance can report on outstanding AR by customer, aging bucket, and invoice type. But invoice records do not link to payment processing events, revenue recognition entries, or usage records. Answering 'has this customer paid and has the revenue been recognized?' requires checking three separate systems.

AI can automate AR aging reports, flag overdue invoices, and detect billing anomalies like duplicate charges. Cannot automate collections optimization or revenue reconciliation because invoice records are isolated from payment and revenue recognition systems.

Invoices are comprehensive records linked to subscriptions, payment transactions, and revenue recognition entries. A controller can query 'show me all invoices over $10K that are past due by more than 30 days for enterprise customers whose usage has declined and whose revenue is still deferred' and get an accurate, actionable result. Invoice dispute records are tracked with resolution history. Credit memos link to the original invoice and adjustment reason.

AI can automate collections prioritization, predict payment timing based on customer payment history, and flag invoices where revenue recognition treatment may need adjustment. Cannot yet autonomously resolve disputes because dispute resolution rules are not machine-readable.

Invoice processing is schema-driven with machine-readable rules. Subscription contracts define billing schedules, proration logic, and tax treatment as structured configuration. Dunning sequences execute based on formalized escalation rules. An AI agent can evaluate 'which past-due invoices should we escalate to account executives versus handle through automated dunning, given the customer's strategic tier, payment history, and renewal timeline?' with a quantified recommendation.

AI can perform fully autonomous invoice management including collections orchestration for routine scenarios. Payment optimization considers customer lifetime value, cash flow timing, and relationship risk. Autonomous AR management is possible for the majority of invoice volume.

Invoices process as a continuous real-time stream. Subscription billing events trigger immediate invoice generation. Payment processing confirmations update invoice status instantly. Dunning sequences execute based on real-time payment signals. There is no 'billing cycle' — invoices flow from generation through collection as a continuous stream with near-zero touch for standard billing scenarios.

Can autonomously manage the complete invoice-to-cash lifecycle in real-time with minimal human intervention for routine billing transactions.

Ceiling of the CMC framework for this dimension.