Build voice-first agents people take everywhere.
Give your agent a voice presence users can discover and talk to by name. Simple text in, text out — Pipbit handles audio, speech, and cellular connectivity.
Voice-native
Users find your agent by speaking. "Hey Pip, connect me with George." No app store, no downloads, no screens. Just a name and a conversation.
Simple webhook model
Your agent is a web service. Pipbit sends user utterances to your webhook, you return a spoken reply. Our Python SDK handles auth, health checks, and delivery in a few lines.
Your users, your auth
Choose how users authenticate — public access, voice credentials, email verification, SMS codes, or full OAuth. You control the trust boundary.
Distinct voice identity
Give your agent a personality. Describe the tone, energy, and tempo you want and we'll match a voice from our catalog. Your agent sounds like itself, not like Pip.
Proactive messaging
Your agent doesn't have to wait to be asked. Push messages to users — reminders, alerts, status updates — and Pip will deliver them naturally in conversation.
AI-assisted development
We provide a development prompt you can hand to Claude or Codex. Describe what your agent should do, and your AI assistant helps build it using the SDK patterns.
from pipbit import Agent
agent = Agent(
name="george",
secret="pb_live_sk_..."
)
@agent.handle
def handle(req):
# req.text is what the user said
# return your agent's spoken reply
return f"Hey! You said: {req.text}"
agent.run(host="0.0.0.0", port=3002)A user speaks to Pip. Pip routes their words to your webhook as text. You return a text reply, and Pip speaks it back. That's the whole contract.