Skip to main content
VAW Native vs Vapi, Retell AI and ElevenLabs

VAW Native Compared with Vapi, Retell AI and ElevenLabs

The main differences between VAW Native and Vapi, Retell AI or ElevenLabs Pods inside VoiceAIWrapper, and how to choose an engine for each client.

Written By voiceaiwrapper

Last updated 26 days ago

VoiceAIWrapper runs agents on two kinds of Pod (an engine or provider connection). VAW Native is the voice engine built into VoiceAIWrapper. Vapi, Retell AI and ElevenLabs are BYOK Pods: you connect your own provider account with its API keys. This article sets out how the two differ inside VoiceAIWrapper, so you can pick the right one for each client.

The comparison covers only what VoiceAIWrapper offers for each engine. Your provider may offer more in its own dashboard. For a feature-by-feature grid, see Provider Feature Comparison.

You don't have to pick one

Every organization has one VAW Native Pod from the start, on every plan. You can connect Vapi, Retell AI or ElevenLabs Pods next to it when your plan includes them, then use whichever engine suits each client. Switch between Pods with the Pod selector in the header.

The header Pod selector open in an organization with a VAW Native Pod and a Vapi Pod: the You are currently in card showing the VAW Native Pod with its provider badge, OUT/IN concurrency pills and Pod ID, and the Vapi Pod listed below it with its โ‹ฎ menu, plus Add Pod at the bottom. Blur Pod names that identify a client.

The differences at a glance

VAW NativeVapi, Retell AI and ElevenLabs Pods
Who you pay for callsVoiceAIWrapper, in AI credits from your organization's balance. Twilio bills you directly for phone numbers and phone minutes.Your provider, on its own bill. Calls don't use AI credits. On some plans, each BYOK Pod also uses a BYOK provider connections add-on.
SetupNone to start. The Pod is already there and needs no API keys. Connect a Twilio account when you want phone calls.Paste your provider's keys, set the Pod's concurrency, and sync each agent and number so call events reach VoiceAIWrapper. ElevenLabs also needs a webhook created in ElevenLabs.
Where agents are built and editedEntirely inside VoiceAIWrapper: from blank, or with VAW Assistant from a business website. Building with VAW Assistant uses AI credits.In your provider's dashboard, or with VAW Assistant on plans that include it. In VoiceAIWrapper you edit the greeting and system prompt.
Phone numbersBought or imported through your Twilio account, inside VoiceAIWrapper.Bought or added in your provider. They appear in VoiceAIWrapper on their own.
Knowledge basesCreated and managed in VoiceAIWrapper from files, web pages and pasted text.Listed in VoiceAIWrapper, read-only. You manage them in your provider.
ToolsBuilt-in End conversation, Transfer call and Send SMS, plus your own custom HTTP and Cal.com tools, all set up in VoiceAIWrapper.Vapi and ElevenLabs: listed read-only, and you can add Cal.com or GoHighLevel booking tools. Retell AI: managed in Retell AI.
Post-call analysisSummary, sentiment and call success written by VoiceAIWrapper, plus Call Data Extractions.Your provider's own call analysis. Call Data Extractions aren't available.
Web widgetVoice and chatVoice and chat. On Retell AI, chat needs a separate Retell AI chat agent.
Client controlsPage by page and setting by setting, with caps such as a maximum number of agents.Access levels per area. For agents: view or edit the greeting and prompt, and allow test calls.
What it costs you per callShown on every call: open it in Call Logs โ†’ What ran โ†’ Cost.Reported by your provider. Campaign Analytics show it as Total Spend.

Web widgets need a plan that includes Web widget. VAW Assistant on a Vapi, Retell AI or ElevenLabs Pod needs a plan that includes VAW Assistant. See Build a Provider Agent with VAW Assistant. On Vapi, Retell AI and ElevenLabs workspaces, client access above View Only needs a plan that includes Advanced client access controls. See Client Controls: What Your Clients Can Do on Each Engine.

How the BYOK Pods differ from each other

VapiRetell AIElevenLabs
Inbound pre-call webhookYesYesNo
Phone number poolsInbound and outbound campaignsInbound and outbound campaignsOutbound campaigns
Extra agent typesWorkflows and SquadsChat AgentsNone
Where the Pod's concurrency comes fromThe Vapi Concurrency Limit you type when you connectRead from your Retell AI accountSet from your ElevenLabs subscription when you connect

VAW Native supports the inbound pre-call webhook and phone number pools for inbound and outbound campaigns. Phone number pools need a plan that includes Phone number pools.

Current limitations

Each engine has gaps inside VoiceAIWrapper. Check these before you promise a feature to a client.

VAW Native

  • No voicemail detection. On outbound calls, a voicemail pickup counts as answered. The campaign setting Consider voicemail as answered isn't offered on VAW Native.
  • No keypad input (DTMF). Callers can't press keys to make choices.
  • Twilio only. Phone numbers must come from a Twilio account you connect.
  • No agent templates or duplicating agents. Each agent starts from VAW Assistant or from blank.
  • One VAW Native Pod per organization.
  • Total concurrency is set by VoiceAIWrapper for your plan. You can change the inbound and outbound split, not the total.

Vapi, Retell AI and ElevenLabs

  • Agents, knowledge bases and most tools are built in the provider. VoiceAIWrapper edits only the greeting and prompt.
  • Webhooks must stay synced. If someone changes an agent's or number's webhook in the provider dashboard, VoiceAIWrapper stops receiving its calls until you click Sync now again.
  • No Call Data Extractions. To save a caller's name and email on the lead, follow the lead capture steps for provider calls instead. See Capture Lead Details from Provider Calls.
  • ElevenLabs has no inbound pre-call webhook in VoiceAIWrapper.

Which should I choose?

Choose VAW Native when:

  • You want one balance in AI credits instead of a separate provider bill for each client.
  • You want to build everything, including knowledge bases and tools, without leaving VoiceAIWrapper.
  • Clients should be able to change more than the prompt, within limits you set.
  • You want to see what every call cost, line by line, before you set client prices.

Choose a Vapi, Retell AI or ElevenLabs Pod when:

  • You or your client already have working agents in that provider.
  • The client wants to own the provider account and pay the provider directly.
  • You rely on something VAW Native doesn't do yet, such as voicemail detection, keypad input or a carrier other than Twilio.
  • You use provider-specific agent types, such as Vapi workflows and squads.

Frequently asked questions

Yes. Each workspace in a client account runs on one Pod, and you can have several Pods. Give each client a workspace on the engine that suits them. See Use Several Providers and Organizations in One Account.

No. Credits are only used on VAW Native. Calls on a BYOK Pod are billed by your provider and never touch your credit balance, so running out of credits doesn't stop them.

Yes, but agents don't move between engines on their own. You rebuild the agent on VAW Native and point the client's campaigns and numbers at it. See Move a Client from Vapi, Retell AI or ElevenLabs to VAW Native.