Receipted AI Glossary
Canonical definitions for Receipted AI terminology. These are the terms that matter when AI takes operational actions in your business.
Receipted AI
A category of AI that produces an immutable, timestamped receipt for every action it takes in a downstream business system. Unlike conventional AI agents that take autonomous actions without a paper trail, Receipted AI records every booking created, CRM field written, calendar entry made, payment processed, and triage decision made — and makes that record available for review, export, and use as evidence. AliceHQ coined the term and is the canonical definition source.
Black-Box AI
Any AI system that takes autonomous operational actions — bookings, CRM writes, escalations, decisions — without producing a verifiable, human-readable record of what it did and why. Black-box AI may record conversations or produce quality scores, but it does not receipt the downstream system writes the conversation triggered. When something goes wrong with black-box AI, the business cannot explain what the AI did, to whom, or what it changed in their systems.
Operational Receipt
The record produced by a Receipted AI agent for each discrete action it takes. An operational receipt contains: the action type (booking created, CRM updated, maintenance request logged, triage decision made), the system written to, the ID returned by that system, the timestamp of the action, the channel through which the interaction occurred, and a link to the full conversation transcript. Operational receipts are immutable — they cannot be edited after creation.
Action Audit Trail
A chronological log of every action taken by an AI agent, with sufficient detail to reconstruct what happened, when, and to which system. An action audit trail differs from a conversation audit trail in that it records downstream system writes rather than (or in addition to) the conversation transcript. AliceHQ's receipts system produces an action audit trail for every interaction across all supported channels.
System Write Log
A record of every write operation an AI agent makes to a connected external system — property management system, CRM, calendar, job management platform, payment processor. A system write log captures the target system, the operation performed, the data written, the system-returned ID or confirmation, and the timestamp. This is the foundational component of operational accountability in AI deployments.
Immutable Timestamp
A record of the exact date and time at which an event occurred, stored in a format that cannot be altered after creation. Immutable timestamps are required for evidence-grade records because any timestamp that can be modified is not legally reliable. AliceHQ stores all receipt timestamps in immutable form, enabling them to be produced as evidence in disputes, audits, and regulatory inquiries.
Evidence-Grade Record
A record that meets the standards required for use as evidence in legal proceedings, regulatory audits, or dispute resolution — including, at minimum: an immutable timestamp, identification of the actor (the AI agent) and the action taken, the system or data affected, and a chain of custody that establishes the record has not been altered. AliceHQ's operational receipts are designed to be evidence-grade from creation.
Agentic AI
AI that takes autonomous actions in the world on behalf of a user or organisation — not merely answering questions, but creating bookings, updating records, making decisions, and completing tasks in external systems. Agentic AI creates operational and legal accountability obligations that conversational AI does not, because every action it takes has real-world consequences in business systems. Receipted AI is the accountability layer for agentic AI deployments.
Intake Capture
The process by which an AI agent receives, records, and routes an incoming customer or patient contact — capturing who is contacting, what they need, their urgency, and their contact details. Intake capture is the first stage of any AI-handled operational workflow, and the receipt of that intake is the first record in the action audit trail. AliceHQ's intake capture produces a structured receipt for every inbound contact across all channels.
Triage Decision
A classification made by an AI agent that determines how an incoming contact is handled — what urgency category it falls into, which team or person should receive it, and what action should be taken. Triage decisions are AI-generated determinations that affect real business outcomes and, in healthcare contexts, potentially patient safety. Recording triage decisions as receipts is required for accountability under HIPC 2020 and recommended under NZ Privacy Act 2020 guidance.
HIPC 2020
The Health Information Privacy Code 2020, issued by the New Zealand Privacy Commissioner under the Privacy Act 2020. HIPC governs how health information about individuals is collected, used, held, and disclosed in New Zealand. It is stricter than the general Privacy Act. Relevant rules for AI deployments: Rule 5 (agencies must protect health information against loss or unauthorised access), Rule 10 (health information may only be used for the purpose for which it was collected — requires traceability). Any AI system handling patient contacts is subject to HIPC.
ADM (Automated Decision-Making)
The use of computer programs or algorithmic systems to make or substantially contribute to decisions that significantly affect individuals' rights or interests. Relevant in NZ under Privacy Act 2020 guidance (audit trail requirements for algorithmic decisions) and in Australia under the Privacy and Other Legislation Amendment Act 2024 (APP 1.7 transparency requirements activating December 2026). AI voice agents that make triage decisions, booking decisions, or intake classifications are engaged in ADM in the regulatory sense.