Pricing

Pricing that scales with every minute.

One unified price per minute. No LLM pass-throughs. No per-token charges.

Starter

$0/mo

$30/month in free credits

Get started

What’s included:

$0.14 / min per conversation

3 Workspace seats

Unlimited voices

10 agents

10 concurrent calls

Pro

$240/mo

$240/month in free credits

Get started

Everything in starter, plus:

$0.12 / min per conversation

30 Workspace seats

50 concurrent calls

50 agents

Support via private Slack Connect

Popular plan

enterprise

Custom Pricing

Customized to your needs

Talk to our team

Everything in pro, plus:

Custom seats

Custom agents

Custom concurrency

On-prem / VPC deployment

Custom terms & assurance for DPAs/SLAs

BAAs for HIPAA compliance

Volume-based discounts

Audit logs

Access to forward deployed engineers

Custom voices

Priority Support

Trusted by leading AI companies

FAQs

Questions you ask before you sign up.

Chat with AI Assistant

Can developers customize agent behavior, conversation flow, and post-call data handling?

Yes. Developers can shape how a Phonic agent behaves by configuring the conversation, passing context, and deciding what happens after the call ends.

Common customization points include:

Agent instructions: define tone, goals, qualification rules, and support boundaries.
Conversation flow: decide what should happen before, during, and after the voice session.
Context injection: pass safe metadata such as userId, accountId, page source, or campaign data.
Post-call handling: send summaries, transcripts, tags, or outcomes to internal systems.
UI behavior: control launch buttons, consent text, loading states, and completion screens.

For editable marketing pages, let Webflow own the visible copy and placement while the server owns session creation and private integration logic.

How should I secure Phonic API keys in a production implementation?

Store Phonic API keys only in a server-side environment, never in Webflow custom code, inline scripts, or public frontend bundles.

A secure production setup should:

Keep the permanent API key in environment variables or a secret manager.
Expose a narrow endpoint such as /api/phonic/session-token.
Validate allowed origins before issuing browser session data.
Return only short-lived credentials to the frontend.
Log failures without exposing secret values.

For general implementation hygiene, the OWASP secrets management guidance is a useful companion reference.

What frontend frameworks can I use with Phonic?

Phonic can be used from most frontend stacks because the browser integration is built around standard web primitives. Teams commonly wire it into Webflow, React, Next.js, or a lightweight vanilla JavaScript embed.

Webflow: good for marketing pages and designer-owned placement.
React or Next.js: good for app surfaces with richer state management.
Plain JavaScript: good for quick embeds and controlled landing pages.

The main requirement is a secure backend path for session creation, plus a frontend surface that can handle microphone permissions and realtime state.

Can Phonic voice agents connect to external tools, webhooks, or internal APIs?

Yes. Phonic agents can be designed to trigger external workflows after or during a conversation, depending on the integration pattern.

Typical developer hooks include:

Webhooks: send conversation outcomes to a CRM, support desk, or data warehouse.
Internal APIs: look up account status, product data, or eligibility rules.
Post-call jobs: summarize transcripts, tag intent, or route follow-up tasks.

For Webflow pages, keep orchestration behind a server endpoint rather than embedding private service credentials in the page.

How does Phonic handle real-time audio streaming in the browser?

Phonic can be implemented as a realtime browser experience: the page captures microphone input, streams audio to the conversation service, and renders agent responses back to the user.

A developer-friendly implementation usually includes:

Browser permission handling for getUserMedia.
A realtime transport such as WebSocket or WebRTC, depending on the integration.
UI states for connecting, listening, speaking, muted, and ended.
Fallback messaging when microphone access is blocked.

Keep the realtime connection details short-lived and request them only when the user starts a conversation.

Can I pass custom metadata, user IDs, or context into a Phonic conversation?

Yes. Pass lightweight metadata at conversation start so downstream systems can connect the conversation to the right user, account, page, or campaign.

Common metadata fields include:

userId or anonymous visitor ID
accountId or workspace ID
sourcePage, campaign, or experiment variant
Any non-sensitive context the agent should use during the session

Do not send passwords, payment data, secrets, or regulated personal data unless the integration has been reviewed for that use case.

For browser-friendly payloads, keep metadata as simple JSON primitives. The MDN JSON guide is a useful reference for the format.

Does Phonic provide an API for generating session tokens or launching conversations?

Yes. In a production setup, the browser should not call Phonic with a permanent secret. Instead, create a small server endpoint that exchanges a protected API key for a short-lived browser session.

Server side: store the API key as an environment variable or managed secret.
Client side: request a temporary session token before opening the voice connection.
Webflow: call only the safe public endpoint, never the secret provider directly.

Example flow: POST /api/phonic/session-token returns the data the frontend needs to launch the conversation. Link this answer to Phonic docs or onboarding materials when the final developer guide is ready.

How do I embed a Phonic voice agent into a Webflow or custom website?

Embed Phonic as a client-side voice experience while keeping sensitive setup on the server. A typical Webflow implementation uses a Code Component or embed for the UI, then calls a backend endpoint to mint a short-lived session token.

Add the Phonic component or embed to the page.
Request a session token from a server route such as /api/phonic/session-token.
Start the browser conversation with the returned realtime connection details.

For the product surface, start at phonic.co. Keep API keys out of Webflow custom code and browser JavaScript.