ElevenLabs is the engine. WFW is the platform.
Every Workforce Wave voice agent runs on ElevenLabs Conversational AI — the same engine powering some of the most advanced voice applications in the world.
WFW adds the infrastructure layer on top: multi-tenancy, vertical intelligence, the /v2/ API, Workforce Wave provisioning, and the policy engine that makes this production-ready.
Workforce Wave Layer
ElevenLabs Engine Layer
Callers and clients interact with WFW. The ElevenLabs engine runs underneath.
The Engine
What ElevenLabs provides.
The three core capabilities WFW runs on top of — provided by ElevenLabs and available in every agent WFW deploys.
ElevenLabs
Conversational AI
ElevenLabs powers the real-time dialogue engine — turn-by-turn conversation with tool-calling, interruption handling, and low-latency response generation.
ElevenLabs
Voice Synthesis
Industry-leading text-to-speech with voice cloning. WFW agents can use ElevenLabs stock voices or custom-cloned voices for brand-consistent audio across every call.
ElevenLabs
Speech Recognition
Real-time speech-to-text transcription with high accuracy across accents and noisy environments — the input layer for every conversation.
The Platform
What WFW adds on top.
ElevenLabs is general-purpose. WFW is the production-ready infrastructure layer that makes it deployable at scale across verticals.
WFW Layer
Multi-Tenancy
One ElevenLabs account powers hundreds of isolated client agents. WFW manages service accounts, access isolation, and per-tenant billing — ElevenLabs has no concept of this.
WFW Layer
Vertical Intelligence Layer
WFW pre-loads industry-specific knowledge (CDT codes, PMS integrations, intake workflows) before the agent is deployed. ElevenLabs is general-purpose — WFW is vertical-specialized.
WFW Layer
Automatic Provisioning
Give WFW a business URL and it generates a configured, deployed ElevenLabs agent in 90 seconds. Without WFW, this requires manual prompt engineering, knowledge base construction, and deployment.
WFW Layer
Bot-Native /v2/ API
The entire WFW REST API surface — agent management, call retrieval, webhook subscriptions, structured extractions — is WFW's layer. ElevenLabs does not expose this to your systems.
WFW Layer
Review Queue and Policy Engine
Every call flows through WFW's review queue, confidence thresholds, and policy enforcement layer before data is surfaced. WFW decides what to flag, escalate, or approve.
WFW Layer
White-Label and Partner Accounts
WFW resellers can deploy branded AI platforms with their own domain, dashboard, and pricing — all powered by ElevenLabs under the hood, invisible to end clients.
How to Think About It
Engine vs. platform — a familiar pattern.
WFW's relationship to ElevenLabs follows the same model as other infrastructure stacks you already use.
| Engine | What it provides | Platform | What it adds |
|---|---|---|---|
| ElevenLabs | Conversational AI, voice synthesis, STT | Workforce Wave | Platform, API, tenancy, integrations |
| Stripe | Payment processing primitives | Shopify Payments | Commerce platform using Stripe as processor |
| AWS | Compute, storage, networking | Vercel | Developer platform running on AWS |
You use Vercel without worrying about AWS EC2 instance sizing. You use WFW without needing to configure ElevenLabs agent prompts, manage API keys per client, or build a review queue from scratch.
The API Layer
The entire /v2/ surface is WFW — not ElevenLabs.
ElevenLabs provides a powerful API for managing individual agents and voices. WFW exposes a completely different API surface designed for multi-tenant platforms, agentic pipelines, and vertical SaaS builders.
- ✓40+ endpoints — agent lifecycle, call management, KB sync, webhooks
- ✓Structured extractions from every call — intent, entities, slots
- ✓Idempotency keys, async operations, scoped service account tokens
- ✓SSE event streams for real-time call monitoring
- ✓MCP server — provision agents from Claude, Cursor, or any MCP host
WFW API — One Call, Deployed Agent
POST /v2/agents
{
"scout_url": "https://smilefamilydental.com",
"vertical": "dental",
"phone_number": "+18435551234",
"voice_id": "eleven_labs_voice_abc"
}
// WFW does automatically:
// 1. Workforce Wave reads the website
// 2. Builds knowledge base
// 3. Generates system prompt
// 4. Creates ElevenLabs agent
// 5. Provisions Twilio number
// 6. Agent live in ~90 seconds
// Returns:
// { "operation_id": "op_abc123", "status": "pending" }ElevenLabs alone requires steps 3 and 4 manually. WFW automates the full stack.
How WFW Uses ElevenLabs
Under the hood — in three steps.
When you provision a WFW agent, here is exactly what happens with ElevenLabs.
Workforce Wave builds the agent config
Workforce Wave reads the business website, generates a system prompt, and builds a knowledge base. This becomes the configuration passed to the ElevenLabs Conversational AI API.
WFW creates the ElevenLabs agent
WFW posts the agent configuration to ElevenLabs under WFW's service account. The ElevenLabs agent ID is stored in the WFW agent record — isolated per tenant.
Calls flow through WFW infrastructure
Inbound calls hit the WFW layer first — Twilio routes to WFW, which connects to ElevenLabs. Transcripts, extractions, and structured data are processed by WFW before being surfaced via the API.
Get Started
ElevenLabs quality. WFW simplicity.
Deploy ElevenLabs-powered voice agents across your entire customer base — without building the infrastructure yourself. Workforce Wave, the /v2/ API, and multi-tenant isolation are all included.