Introduction
Phonic is the platform for building, evaluating, and deploying voice AI agents that sound like a real conversation. Our state-of-the-art speech models feel genuinely human with real emotion, timing, and nuance.
Every agent you build gets:
- Sub-500ms latency between speech-in and speech-out
- Natural interruption handling for real back-and-forth conversation
- Accurate pronunciation of URLs, emails, and alphanumerics
- Enterprise readiness with HIPAA & SOC2 compliance, backed by a 99.9% uptime SLA
- Usage-based pricing starting at $0.15 per minute of conversation
Get started
To start building, you need a Phonic account and an API key.
- Phonic account — sign in or sign up at phonic.ai
- API key — create one on the API keys page and store it in
.env.local
Build
See the API reference for the full surface.
Install the SDK and make your first call.
Install the SDK and make your first call.
Drive the API from your terminal.
Build with AI
Point your AI assistant at https://docs.phonic.ai/docs/get-started/overview.md to have it start building for you.
Every docs page is accessible as clean markdown for token-efficient parsing — append .md to any URL.
Connect Claude Code, Cursor, or Codex to manage your workspace.
Give your coding agent specialized Phonic knowledge.
Point your agent at a full index of these docs.
Example use cases
Phonic agents are already handling conversations like these in production:
Field support inquiries — order or booking changes, refunds, account changes, and escalation options.
Conduct structured real-time voice interviews, evaluates candidates against role fit, and fits into hiring workflows.
Conduct qualitative user interviews with open-ended discovery, neutral probing, and no leading.
Run patient intake, confirms or reschedules appointments, and hands urgent cases off to a nurse line.
Next Steps
Want to go deeper? Start here:
- Conversations — compare webhook and WebSocket conversations
- Tools — learn about the tool types available on Phonic
- Prompting guide — write effective prompts for voice agents