Clonvo AI-First Unified Inbox

The AI tier — grounded, cited answers

• Retrieval-augmented replies that cite the exact document chunks used
• If no answer scores above the confidence threshold, the agent escalates instead of guessing
• Per-business persona, tone, and do-not-reply rules; always identifies itself as AI when asked
• Token spend visible per conversation

The human tier — one-tap hand-off

• When the AI shouldn't answer, it pauses and pings the team with full context
• The human sees customer history, the AI's last reply, and the reason for escalation
• Per-conversation AI/human toggle, internal notes, SLA timers and unread surfacing

Control room & operations

• One live inbox for AI replies, human takeovers, and escalations across all channels
• Hour-by-hour volume with median reply time and per-channel breakdown
• Urdu, Hindi, and English transcription; per-business cost and latency tracking

Onboarding, security & stack

• Zero-to-live in three steps: provision workspace, connect Meta channels, upload knowledge and ship
• Multi-tenant with AES-256 secrets, role-scoped access, and Clerk-based team invites
• Operates safely inside Meta's 24-hour policy windows with compliant templating