Period-End Close Process

The close process lives entirely in the controller's head. When asked 'what's left to do before we can close the books?' the answer is a verbal checklist that changes depending on what the controller remembers. There's no written task list, no documented dependencies, and no defined timeline. Close takes as long as it takes, and nobody knows whether it's on track until it's done — or not.

AI cannot assist with close management because no process definition exists. Cannot identify bottlenecks, predict completion, or automate any close tasks.

A close checklist exists in an Excel spreadsheet — a list of 40-60 tasks with columns for owner, due date, and status. The controller updates it manually during close, checking off tasks as they're completed. But dependencies aren't documented — when AP reconciliation is late, nobody can see which downstream tasks are now blocked. The checklist from last quarter gets copied and modified each period.

AI could read the checklist but cannot track progress in real-time because updates depend on the controller remembering to update the spreadsheet. Cannot identify the critical path or predict close completion because dependencies aren't documented.

The close process follows a documented workflow with task dependencies. Each task has a defined owner, predecessor tasks, target duration, and sign-off requirement. The close calendar shows the critical path and identifies which tasks are blocking. Reconciliation requirements are documented for each account with materiality thresholds. But the workflow is in a spreadsheet or document — it doesn't connect to the actual task execution in the ERP.

AI can analyze the close workflow for optimization opportunities and predict completion based on the documented schedule. Cannot track actual progress because the workflow document doesn't connect to task execution — status updates are manual.

The close process is managed in a dedicated close management tool with real-time task tracking. Each task is linked to its predecessor and successor, assigned to a specific person, and tracked against a target completion time. When the AP team completes their reconciliation, the system automatically notifies the next task owner. Progress dashboards show the critical path, overdue tasks, and predicted close completion date. Journal entry review workflows are system-enforced.

AI can predict close completion based on real-time task progress, identify bottlenecks as they emerge, and recommend task reprioritization. Cannot yet automate the execution of close tasks themselves — only the tracking and coordination.

The close process is fully schema-driven with formal entity relationships. Each close task links to its data sources, required reconciliation steps, journal entry templates, sign-off authorities, and quality gates. An AI agent can query 'what are all uncompleted tasks blocking the revenue close, who owns them, and what data do they need?' and get a structured, actionable answer. Task definitions include machine-readable execution instructions, not just descriptions.

AI can autonomously execute routine close tasks — generate reconciliations, prepare standard journal entries, validate balances — and coordinate the workflow. Human involvement required only for judgment-intensive tasks and final sign-off.

The close process is continuous and self-executing. Reconciliations run as transactions post, journal entries generate from transaction patterns, and exception reviews trigger as anomalies are detected — not at period-end. The 'close' is a final validation step, not a multi-day work marathon. The system documents itself — close status, task completion, and sign-off trails generate from process execution without separate tracking.

Fully autonomous close management. AI executes, tracks, validates, and reports the close process with minimal human intervention. Period-end close is reduced from days to hours.

Ceiling of the CMC framework for this dimension.