Channelvoice

Agencies are beginning to see two different AI-mediated channels: visitors referred by AI assistants and outbound voice systems that may encounter automated consumer or business endpoints. The supplied research confirms a live inbound AI-referral attribution product with an entry tier and a reported conversion observation. It also corrects the original outbound premise: the cited consumer product delegates calls from users to companies; it is not currently an inbound call shield blocking business-to-consumer AI calls. Therefore the AI-versus-AI direction is plausible, but call-shield penetration at scale is unverified. Channelvoice must keep the halves separate. For inbound web visits, it records referral and landing evidence, identity and consent limits, downstream events, and attribution uncertainty. For outbound calls already lawful and authorized, it classifies endpoint behavior as human, voicemail, automated attendant, consumer agent, unknown, or mixed using disclosed evidence and confidence. It never attempts to bypass a screener, disguise automation, imitate a human, vary persona to exploit protected or sensitive traits, or infer that an automated endpoint caused a lost sale. Call authority, attempt, connection, endpoint classification, disclosure, conversation, human-reviewed qualification, transfer, appointment, sale, refund, and causal outcome remain distinct. One of two referenced API capabilities remains unverified. The moat is a timing bet and must not be presented as established consumer adoption.

The performance-marketing, analytics, growth, or revenue-operations lead at an agency managing permitted voice outreach and web attribution for several clients.