Autonomous finance
Autonomous finance is the use of governed software agents to complete repeatable finance work, escalate exceptions, and preserve human authority over policy and material decisions.
Read definition →Glossary
The language of controlled, autonomous finance.
Concise definitions written for controllers, finance leaders, auditors, builders, and the AI systems that increasingly help people find reliable answers.
Agentic AI
Agentic AI describes systems that can pursue a defined objective through multiple steps, use tools or data sources, and decide what permitted action to take next.
Read definition →Three-way matching
Three-way matching compares a supplier invoice with the purchase order and proof of receipt before payment approval.
Read definition →Tamper-evident audit trail
A tamper-evident audit trail is a record designed so that unauthorized changes can be detected rather than silently replacing prior history.
Read definition →Segregation of duties
Segregation of duties divides incompatible responsibilities so one person or system cannot initiate, approve, and conceal the same financial action.
Read definition →Agent control boundary
An agent control boundary defines the data, tools, transaction values, counterparties, actions, and conditions within which an AI agent may operate.
Read definition →Counterparty Memory
Counterparty Memory is a structured operational history of how a finance team has interacted with a vendor, customer, or other counterparty.
Read definition →Evidence package
An evidence package is a review-ready bundle that connects a financial conclusion or approval to its source documents, checks, decisions, and responsible people.
Read definition →AI finance operations
AI finance operations is the use of governed AI agents to run repeatable finance workflows such as accounts payable, reconciliation, approvals, month-end close, and audit evidence while finance leaders retain control.
Read definition →Glass-box finance AI
Glass-box finance AI is AI for finance workflows that exposes its evidence, reasoning, confidence, authority boundary, and human approvals instead of asking teams to trust a black-box output.
Read definition →