End-to-end build orchestrator for Agentforce Contact Center (AFCC) with native Salesforce Voice. Use when standing up a comprehensive AFCC voice agent from scratch — sequencing licenses, the Voice channel, Agentforce Voice, Agentforce for Service, Command Center, and advanced IVR/callbacks — and to decide which focused AFCC skill to hand off to next.
インストール
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
End-to-end build orchestrator for Agentforce Contact Center (AFCC) with native Salesforce Voice. Use when standing up a comprehensive AFCC voice agent from scratch — sequencing licenses, the Voice channel, Agentforce Voice, Agentforce for Service, Command Center, and advanced IVR/callbacks — and to decide which focused AFCC skill to hand off to next.
disable-model-invocation
true
AFCC Build Orchestrator
Use This Skill When
You are building a full Agentforce Contact Center with native Salesforce Voice and need the correct end-to-end sequence.
You need to decide which focused AFCC skill applies to the current step (channel, voice agent, IVR, media, callbacks, Agentforce for Service, Command Center).
You are validating readiness at each phase gate before moving on (dial-ready, agent-ready, escalation-ready, supervisor-ready).
You want a single map of prerequisites, dependencies, and demo checkpoints for AFCC.
Prerequisites
A Salesforce demo org (SDO) provisioned Feb 21, 2026 or later on a NA/AMER instance. Use a Spring release org until SDOs are approved for Summer.
Admin user assigned the AFCC permission sets (see sf-afcc-voice-channel-setup).
This scope is native Voice only — do not combine with Salesforce Voice + Amazon Connect / BYOT in the same org.
Core Workflow
Phase 1 — Voice channel foundation → hand off to sf-afcc-voice-channel-setup
Assign permission sets, claim a phone number, configure and activate the Voice channel, set CLI, enable recording/transcription, set routing config, build the Voice Call record page.
Gate: place a live inbound test call that routes to your agent in the Service Console.
Phase 2 — Data 360 enablement → prerequisite for Agentforce Voice
Enable OAuth username-password flows, then run the Data 360 setup automation (Q-Brix) or manual Data Cloud setup. Activation takes ~60 minutes.
Gate: Data Cloud setup reports success.
Phase 3 — Agentforce Voice agent → hand off to sf-afcc-agentforce-voice-agent
Create/enable the Agentforce Service Agent, connect Telephony, build the inbound routing flow and escalation flow, tune voice settings, activate.
Gate: an inbound call is handled by Agentforce Voice and escalates cleanly to a human rep with full context (single Voice Call record, split recordings).
Phase 4 — IVR & media (optional/advanced) → hand off to sf-afcc-ivr-flow and sf-afcc-media-management
Add Play Prompt IVR (DTMF, SSML, voice profiles), manage audio media, set recording/transcription behavior, add End Call.
Gate: DTMF menu branches correctly; prompts play with the intended voice/media.
Phase 5 — Callbacks (advanced) → hand off to sf-afcc-callbacks
Configure the Contact Request object, callback queue, and callback flow with auto-dial.
Gate: a requested callback routes to a rep and auto-dials a US number.
Phase 6 — Agentforce for Service → hand off to sf-afcc-agentforce-for-service
Enable Conversation Catch-Up and Work Summaries for Voice; map summary fields.
Gate: a call with 10+ transcript entries generates a summary.
Phase 7 — Command Center (Supervisor) → hand off to sf-afcc-command-center
Configure Supervisor tabs, visible Agentforce agents, and Service reports/dashboards.
Gate: supervisor can monitor queues, reps, and Agentforce Voice sessions.
Throughout — Troubleshooting → consult sf-afcc-troubleshooting
Follow the phase order — Agentforce Voice depends on Data 360, and escalation depends on an activated channel and correct queue routing config.
Consider the Q-Brix automation ("Salesforce Voice | SDO Q Brix") to get dial-ready fast, but do a manual pass at least once so you can credibly answer "how hard is this to set up?"
Deactivate Voice Call On Create automations that create cases/log calls before relying on outbound/callback flows — they commonly cause stuck "dialing" and multi-call connect issues.
Keep one Voice Call record mental model: native AFCC re-queues the same record on escalation (unlike Amazon/BYOT which creates two).
Deliverables
A phased AFCC build plan with per-phase gates and the specific skill to execute each phase.
A prerequisite and dependency checklist (org age/instance, licenses, permissions, Data 360).
A demo-readiness checklist mapped to capabilities to show (softphone, transcription, escalation, summaries, supervisor).
