Build a Metered AI Voice Agent with Clerk Billing + Stripe (2026)

TL;DR — Clerk Billing (0.7% + Stripe fees) gives you a drop-in <PricingTable />, server-side has({ feature }) entitlement, and Stripe Meters for per-minute voice billing — without writing a webhook handler.

What you'll build

Three voice plans (Starter / Pro / Scale), each gating minutes per month. The Realtime endpoint checks entitlement, increments a Stripe Meter on every minute, and shuts off cleanly on cap.

Prerequisites

1. Next.js 15 + Clerk @clerk/nextjs@^6.
2. Stripe account connected to Clerk Billing.
3. OPENAI_API_KEY, STRIPE_SECRET_KEY.

Architecture

flowchart TD
  U[User] --> CL[Clerk Auth + Plan]
  U --> RT[/api/realtime/]
  RT --> EN{has voice_minutes?}
  EN -->|yes| OA[OpenAI Realtime]
  OA --> M[Stripe Meter increment]
  EN -->|no| UP[Upgrade page]

Step 1 — Define plans in Clerk Dashboard

In Clerk Dashboard → Billing, create plans starter_149, pro_499, scale_1499 and a feature voice_minutes. Mark voice_minutes as a metered feature linked to a Stripe Meter voice_minutes_meter.

Step 2 — Drop the pricing table

```tsx // app/pricing/page.tsx import { PricingTable } from "@clerk/nextjs"; export default function Pricing() { return ; } ```

Step 3 — Entitlement gate

```ts // app/api/realtime/route.ts import { auth } from "@clerk/nextjs/server"; import { NextResponse } from "next/server";

export async function POST() { const { userId, has } = await auth(); if (!userId) return new NextResponse("auth", { status: 401 }); if (!has({ feature: "voice_minutes" })) return NextResponse.json({ upgrade: true }, { status: 402 });

const r = await fetch("https://api.openai.com/v1/realtime/sessions", { method: "POST", headers: { Authorization: Bearer ${process.env.OPENAI_API_KEY}, "Content-Type": "application/json" }, body: JSON.stringify({ model: "gpt-realtime" }), }); return NextResponse.json(await r.json()); } ```

Step 4 — Meter usage on call end

```ts import Stripe from "stripe"; import { auth, clerkClient } from "@clerk/nextjs/server";

const stripe = new Stripe(process.env.STRIPE_SECRET_KEY!);

export async function POST(req: Request) { const { userId } = await auth(); const { minutes } = await req.json(); const u = await (await clerkClient()).users.getUser(userId!); const customerId = u.publicMetadata.stripeCustomerId as string;

await stripe.billing.meterEvents.create({ event_name: "voice_minutes_meter", payload: { stripe_customer_id: customerId, value: String(minutes) }, }); return Response.json({ ok: true }); } ```

Step 5 — Hook on call disconnect

In the browser, when the WebRTC peer goes disconnected, POST elapsed minutes to /api/meter.

Step 6 — Cap enforcement

Use has({ feature: "voice_minutes", quantity: { gte: capLeft } }) in middleware to block once the soft cap is hit.

Pitfalls

• Test mode: Clerk Billing dev sandbox uses Stripe test keys — switch both before launch.
• Meter granularity: Stripe Meters dedupe by (customer, event_name, idempotency_key, time); always pass a unique identifier per call.
• Entitlement caching: has() is cached for 30s — for hard caps, query Stripe usage directly.

How CallSphere does this in production

CallSphere prices at $149 Starter / $499 Pro / $1,499 Scale with metered overage, 14-day no-card trial, and 22% recurring Year-1 affiliate. The platform spans 37 agents, 90+ tools, 115+ DB tables, 6 verticals — Healthcare, OneRoof (Next.js 16 + React 19), Salon (NestJS 10 + Prisma), Sales (Node.js 20 + React 18 + Vite). Clerk + Stripe handles auth + billing across all six.

FAQ

Clerk Billing fee? 0.7% per transaction on top of Stripe fees.

Can I run B2B teams? Yes — Clerk has Organizations with seat-based plans and per-org features.

Can I avoid Clerk and use Stripe Customer Portal directly? Yes, but you give up the React components and has() helpers.

EU compliance? Clerk offers SCA-ready Stripe checkout out of the box.

Sources

• Clerk Billing for B2C - https://clerk.com/docs/nextjs/guides/billing/for-b2c
• Clerk Billing for B2B - https://clerk.com/docs/nextjs/guides/billing/for-b2b
• Stripe Sessions: Clerk + Stripe - https://stripe.com/sessions/2025/instant-zero-integration-saas-billing-with-clerk-stripe
• Stripe Billing Meters - https://docs.stripe.com/billing/subscriptions/usage-based/recording-usage

How this plays out in production

One layer below what Build a Metered AI Voice Agent with Clerk Billing + Stripe (2026) covers, the practical question every team hits is multi-turn handoffs between specialist agents without losing slot state, sentiment, or escalation context. Treat this as a voice-first system from the first prompt: the agent's persona, its tool surface, and its escalation rules all flow from that single decision. Teams that ship fast tend to instrument the loop end-to-end before they tune any single component, because the bottleneck is rarely where intuition puts it.

Voice agent architecture, end to end

A production-grade voice stack at CallSphere stitches Twilio Programmable Voice (PSTN ingress, TwiML, bidirectional Media Streams) to a realtime reasoning layer — typically OpenAI Realtime or ElevenLabs Conversational AI — with sub-second response as a hard SLO. Anything north of one second of perceived silence and callers either repeat themselves or hang up; that single number drives the whole architecture. Server-side VAD with proper barge-in support is non-negotiable, otherwise the agent talks over the caller and the conversation collapses. Streaming TTS with phoneme-aligned interruption keeps the cadence natural even when the user changes their mind mid-sentence. Post-call, every transcript is run through a structured pipeline: sentiment, intent classification, lead score, escalation flag, and a normalized slot extraction (name, callback number, reason, urgency). For healthcare workloads, the BAA-covered storage path, audit logs, encryption-at-rest, and PHI-safe transcript redaction are wired in from day one, not bolted on at compliance review. The end state is a system where every call produces a row of structured data, not just a recording.

FAQ

What is the fastest path to a voice agent the way Build a Metered AI Voice Agent with Clerk Billing + Stripe (2026) describes?

Treat the architecture in this post as a starting point and instrument it before you tune it. The metrics that matter most early on are end-to-end latency (target < 1s for voice, < 3s for chat), barge-in correctness, tool-call success rate, and post-conversation lead score distribution. Optimize whatever the data flags as the bottleneck, not whatever feels slowest in your head.

What are the gotchas around voice agent deployments at scale?

The two failure modes that bite hardest are silent context loss across multi-turn handoffs and tool calls that succeed in dev but get rate-limited in production. Both are solvable with a proper agent backplane that pins state to a session ID, retries with backoff, and writes every tool invocation to an audit log you can replay.

What does the CallSphere outbound sales calling product do that a regular dialer does not?

It uses the ElevenLabs "Sarah" voice, runs up to 5 concurrent outbound calls per operator, and ships with a browser-based dialer that transfers warm calls back to a human in one click. Dispositions, transcripts, and lead scores write back to the CRM automatically.

See it live

Book a 30-minute working session at calendly.com/sagar-callsphere/new-meeting and bring a real call flow — we will walk it through the live outbound sales dialer at sales.callsphere.tech and show you exactly where the production wiring sits.