
Phonic
Speech-to-speech model for voice agents that can call tools while the caller is still talking.

About Phonic
Phonic: Speech-to-Speech Model for Voice Agents
Phonic builds Phonic-1, a speech-to-speech model for voice agents that hears the caller's audio directly instead of a transcript. It picks up tone, pace and hesitation, responds in about 600 ms and can run tool calls while the caller is still speaking. Developers reach it through one SDK and API, billed at a single price per conversation minute.
Key Features
Speech-to-Speech Model: Phonic-1 processes audio end to end, replacing the usual chain of separate speech-to-text, LLM and text-to-speech models.
Tool Calls While Talking: The model can call tools and take action while the caller is still speaking, so the answer is ready when they stop instead of a "let me check on that" pause.
Supervisor Models: Frontier models act as supervisor agents to assist Phonic-1 with tool calling and reasoning on complex tasks.
Response Time: Phonic states a response time of about 600 ms, with turn-taking tuned for noisy lines.
Multilingual: Trained on 50+ languages and able to switch language mid-conversation while keeping the same speaker identity.
Integrations: Plugs into existing stacks through webhooks, a held-open websocket or LiveKit, without rebuilding current infrastructure.
Deployment Options: Multi-tenant Phonic Cloud by default, plus single-tenant, VPC peering and self-hosting for Enterprise customers.
Security: SOC 2 Type II certified, HIPAA with BAA available, SSO (SAML / OIDC) and a 99.99% uptime SLA.
Use Cases
Customer Support: Agents that handle long, multi-step calls with many tool calls and natural interruption handling.
Recruiting: Voice screening and AI interviews for candidates in multiple languages.
Healthcare: Patient intake calls, with HIPAA coverage and a BAA available.
Logistics: Dispatch coordination and operational calls that need to look up and update data mid-call.
Pricing
Starter: $0.07 per minute, no minimum, 10 concurrent calls and 10 agents.
Pro: $0.065 per minute with a $200 monthly minimum, 50 concurrent calls and 50 agents.
Enterprise: Custom pricing with volume discounts and forward deployed engineers.
Getting Started
Website: https://phonic.ai
Documentation: https://docs.phonic.ai
Phonic gives teams building voice agents a single audio-native model in place of a multi-vendor STT, LLM and TTS pipeline, aimed at complex calls that need reasoning, tool use and natural delivery.
Funding
$4M raised ยท 2 rounds

