Millis AI is a voice AI platform founded in 2024 by Thach Nguyen and Shang-Pin Sheng, headquartered in San Leandro, California. It enables developers and non-technical users to build and deploy conversational voice agents with an advertised end-to-end latency of 600ms, which the company positions as among the fastest available for voice AI pipelines. The platform supports both a no-code visual builder and a code-based SDK path, allowing users to go from prompt to deployed agent in roughly five minutes. Agents can be embedded into web apps via a widget, deployed to web, mobile, or desktop via SDKs, or connected to phone numbers for inbound and outbound telephony across 100+ countries. Millis AI is designed to be provider-agnostic at each layer of the voice stack. Users can connect their choice of LLM (OpenAI, Mistral, Llama, or a custom endpoint), text-to-speech provider (ElevenLabs, PlayHT, Cartesia, or a cloned voice), and speech-to-text provider. The platform charges a base rate on top of these third-party provider costs rather than bundling them, giving teams flexibility to optimize for cost or quality at each layer. The primary use cases are customer support automation, outbound sales and lead qualification, appointment scheduling, and always-on virtual assistants. The company is bootstrapped with no external funding reported, and reached approximately $330K in annual revenue by 2025 with a team of three people. Key features: - Ultra-low 600ms end-to-end voice latency - No-code builder plus SDK/API for developers - Global telephony for inbound and outbound calls in 100+ countries - Provider-agnostic LLM, TTS, and STT integrations (OpenAI, ElevenLabs, Cartesia, PlayHT, Mistral, Llama, custom) - Web widget embed and multi-platform SDK deployment - API integrations to connect agents with CRMs, calendars, and SaaS tools - Custom voice cloning support - Usage-based pricing with volume discounts
$0.02/minute base platform fee, plus third-party LLM/TTS/STT costs; usage-based with volume discounts; enterprise pricing available on request