Estuary Ai

Estuary is a platform designed for building AI characters that can see, speak, and remember, emphasizing real-time interaction and persistent memory across sessions and platforms. It addresses the challenges developers face when creating socially interactive agents, such as fragmented APIs, session-based memory loss, and the need for custom infrastructure. Estuary enables developers to create characters that recall information across conversations and devices, supporting memory that compounds over time through a knowledge graph built from every interaction.

The platform offers a modular architecture for multimodal AI characters, integrating capabilities for text, audio, and vision processing. Its persistent memory system allows characters to remember facts, preferences, relationships, events, emotions, corrections, topics, and entities, enabling richer and more personalized interactions. Estuary supports multi-character deployments, where several AI agents can share context, converse with each other, and coordinate autonomously. The voice engine provides per-character voices, allowing developers to clone voices from samples, select from a library, or use custom voices, with built-in support for natural interruption and turn-taking. Developers can mix and match speech-to-text (STT), language model (LLM), and text-to-speech (TTS) providers for each character, optimizing for cost, latency, or quality.

Character creation is facilitated through a unified dashboard, where developers can define personality, craft voices, and configure model stacks. The platform offers AI-assisted persona authoring, voice cloning, and vendor-agnostic configuration. Estuary is designed for cross-platform deployment, supporting web, Unity, Python, Lens Studio, npm, mobile, desktop, and smartglasses, allowing characters to follow users across devices and retain memory throughout. It supports real-time interaction with low latency and can run entirely off-cloud for data privacy.

Estuary has been used in academic and hackathon projects, including AI toys on Raspberry Pi, smartglass assistants, and multiuser AI pet experiences, demonstrating its flexibility and applicability in various contexts. The platform is referenced in published research as a framework for building multimodal, low-latency, real-time socially interactive agents. Estuary is developed by Estuary Systems, Inc.