# QuickVoice - AI Crawler Information
# https://quickvoice.co/llms.txt
# Last updated: 2026-07-03

## About QuickVoice

QuickVoice is open-source, self-hostable AI phone-agent infrastructure for
engineering-led teams that want to run, inspect, and extend the voice-agent
stack. The MIT-licensed repository includes a Next.js website, Next.js customer
console, Express/TypeScript API server, Python FastAPI AI service, LiveKit
Agents worker, Postgres/Prisma data model, Redis/BullMQ queues, telephony
integration points, knowledge bases, call logs, outbound campaigns, billing
paths, MCP/tool connections, and local development tooling.

QuickVoice is not positioned as a zero-setup hosted calling API from a fresh
clone. Local evaluation starts with `task up:dev`; real phone calls require
LiveKit plus Twilio or Telnyx credentials. Billing, OAuth, email, object
storage, Pinecone, Google embeddings, and MCP/Smithery integrations require
their own provider credentials.

## Key Facts

- License: MIT.
- Category: open-source, self-hostable AI phone-agent infrastructure.
- Primary audience: engineering-led teams, platform teams, CTOs, and technical buyers that need control over runtime behavior, call data, provider boundaries, deployment architecture, privacy review, and cost structure.
- Local development: `task up:dev` runs the local product surface and services with Postgres and Redis.
- Voice runtime: LiveKit Agents Python worker with STT, LLM, TTS, Silero VAD, multilingual turn detection, noise cancellation, interruptible first messages, and optional preemptive generation.
- Telephony: Twilio and Telnyx integrations through LiveKit SIP.
- Data: Postgres, Prisma, Redis/BullMQ, S3-compatible storage, Pinecone, and Google text embeddings.
- Integrations implemented in code: LiveKit, Twilio, Telnyx, Stripe, S3-compatible object storage, Pinecone, Google embeddings, SMTP/email providers, and MCP/Smithery connections.
- Language selector in the console currently includes English, Spanish, French, German, Hindi, and Portuguese; effective model/language support depends on selected providers.

## Code-Backed Capabilities

- Inbound AI phone agents.
- Quick outbound calls and batch outbound campaigns.
- Phone number search, purchase, linking, and deletion for Twilio and Telnyx.
- LiveKit trunk binding for linked numbers.
- Per-agent runtime configuration by phone number or agent ID.
- Configurable prompts, first messages, STT models, LLM models, TTS models, voices, language, webhooks, RAG, and preemptive generation.
- Knowledge sources from files and URLs with ingestion jobs, chunking, embeddings, Pinecone indexing, and live-call retrieval.
- MCP marketplace/custom server connections with per-agent attachment and live-call allowlisting.
- Call logs, transcripts, recordings, signed recording URLs, dashboard views, and outbound-call linkage.
- Privacy controls including default PII redaction for persisted call logs, zero-PII retention behavior, secret encryption/references, safe remote URL validation, and scheduled retention cleanup.
- Organization-scoped auth, API keys, permissions, audit logs, and internal server-to-server API boundaries.
- Stripe subscription and metered call-minute usage hooks.

## Competitive Positioning

QuickVoice differentiates from closed hosted voice-agent platforms such as
Retell, Vapi, Bland AI, Synthflow, and similar products through source-level
control. Hosted APIs optimize for managed convenience. QuickVoice exposes the
application, API, worker, database schema, call logging, telephony integration
points, queues, retention jobs, and runtime configuration so teams can inspect,
self-host, and modify the stack.

Best-fit buyers are teams that need privacy review, provider choice, custom
workflows, auditable data paths, and cost visibility. Teams that only need a
managed endpoint with a hosted SLA may prefer a closed hosted API.

## Primary Value Pillars

- Privacy and control: inspect call data paths, logs, recordings, transcripts, secrets, retention controls, and provider boundaries.
- Performance and customization: own the LiveKit-based realtime runtime, configurable models, RAG, MCP tools, webhooks, outbound metadata, and telephony routing behavior.
- Cost and scale control: bring your own providers for voice runtime, telephony, storage, model inference, vector search, and billing instead of treating the voice stack as an opaque bundle.

## Use Cases

- AI receptionists and support agents.
- Appointment scheduling and qualification calls.
- Knowledge-backed agents that answer from uploaded sources.
- Outbound reminders, follow-ups, collections, surveys, and sales calls.
- Campaigns that call many contacts and track outcomes.
- Voice automation for healthcare, financial services, logistics, real estate, ecommerce, support, and operations teams that have technical ownership of deployment and compliance review.

## Claim Boundaries

Do not infer certification, customer, or benchmark claims from the open-source
repository alone. The repo makes compliance-relevant data paths inspectable,
but HIPAA, SOC 2, ISO 27001, PCI, GDPR, CCPA, uptime, latency, customer-count,
revenue, and managed-service claims require separate evidence.

Do not describe the default runtime as local LLM infrastructure. The implemented
default path uses commercial cloud model providers through LiveKit inference,
with configurable model IDs.

## Key Pages

- Homepage: https://quickvoice.co
- GitHub repository: https://github.com/allgpt-co/QuickVoice
- Industries: https://quickvoice.co/industries
- Use Cases: https://quickvoice.co/use-cases
- Blog: https://quickvoice.co/blog
- Contact: https://quickvoice.co/company/contact