Production-Ready ADK

Unite and Orchestrate
AI Agents with

AgentSea Logo

The complete framework for building agentic AI applications with built-in voice, commerce capabilities, local models, and complete privacy.

12+
LLM Providers
15+
Packages
7
Voice Providers
100%
TypeScript
02

Voice-Enabled AI in 5 Lines

AgentSea is the only framework with built-in voice support. Add STT/TTS to your agents instantly.

import { VoiceAgent, OpenAIWhisperProvider, OpenAITTSProvider } from '@lov3kaizen/agentsea-core';
const voiceAgent = new VoiceAgent(agent, {
  sttProvider: new OpenAIWhisperProvider(),
  ttsProvider: new OpenAITTSProvider({ voice: 'nova' }),
});
// Voice in → Voice out
const result = await voiceAgent.processVoice(audioFile, context);
04

Everything You Need

A comprehensive ADK with batteries included

🎙️

Voice Support (TTS/STT)

7 voice providers built-in: OpenAI Whisper, LemonFox, Local Whisper, OpenAI TTS, LemonFox TTS, ElevenLabs, Piper TTS

🏠

Local Models

6 local providers: Ollama, LM Studio, LocalAI, Text Generation WebUI, vLLM, and more

🛠️

CLI Tool

Interactive setup, chat, agent management, and model management in one command

👥

Multi-Agent Crews

Role-based coordination with delegation strategies and workflow management

12+ Providers

Support for Anthropic Claude, OpenAI GPT, Google Gemini, and 9 more providers

🔌

MCP Protocol

First-class Model Context Protocol support with built-in Figma & n8n tools

🎨

Figma & n8n Integration

Built-in tools for Figma designs and n8n workflow automation with retry logic

🛒

ACP Commerce Protocol

14 commerce tools for product search, cart management, checkout, and secure payments

Content Formatting

Format responses as text, with syntax highlighting and theming

💾

Advanced Memory

Buffer, Redis, and summary memory stores for context management

🚀

LLM Gateway

Intelligent routing with load balancing, failover, and cost optimization

🌐

REST API & Streaming

Built-in HTTP endpoints, Server-Sent Events (SSE), and WebSocket support

🏗️

NestJS Ready

First-class NestJS integration with decorators and modules

🛡️

Safety Guardrails

Content filtering, PII detection, and prompt injection protection

📊

LLM Evaluation

Automated metrics, LLM-as-Judge, and human feedback collection

🧬

Vector Embeddings

Multi-provider embeddings with chunking and vector store integrations

🌊

Browser Automation

Vision-based browser control and computer use with Claude vision

🔒

Complete Privacy

Run everything locally - LLM, voice processing, and data storage

🏢

Multi-Tenancy

Enterprise-grade tenant isolation, API key auth, quota management

05

Run Everything Locally

Run LLMs and voice processing completely on your hardware. No cloud, no API costs, complete data control.
npm install @lov3kaizen/agentsea-core
Everything runs locally - complete privacy
const privateAgent = new VoiceAgent(
  new Agent({ model: 'llama2', provider: 'ollama' }),
  { sttProvider: new LocalWhisperProvider() }
);
06

Why Choose AgentSea?

Built for developers who value simplicity and completeness

vs LangChain

  • +Voice built-in (7 providers)
  • +TypeScript-first (not a port)
  • +CLI tool included
  • +Local models prioritised

vs Building from Scratch

  • +Setup in 15 minutes (vs 1-2 days)
  • +Save $75K-150K in Year 1
  • +Production ready Day 1
  • +Comprehensive docs included

Ready to Build?

Start creating powerful voice-enabled AI agents today