Intelligent gateway for agents

Go beyond nascent AI demos

Effortlessly build AI apps that can answer questions and help users get things done. Arch is the AI-native proxy that handles the pesky heavy-lifting so that you can move faster in building agentic apps, prevent harmful outcomes, and rapidly incorporate latest models.

Why Arch Gateway

Built by the contributors of Envoy Proxy with the belief that:

Prompts are nuanced and opaque user requests, which require the same capabilities as traditional HTTP requests including secure handling, intelligent routing, robust observability, and integration with backend (APIs and tools) to build agentic tasks – outside business logic.

Benefits

What You Can Achieve

Launch Faster

Focus on core objectives

Arch is integrated with purpose-built LLMs to handle the critical but pesky heavy lifting in building agentic apps. Arch offers fast request clarification, query routing, and data extraction from user requests so that you can move faster in building enterprise-worthy agentic apps -- without the taxing prompt engineering and systems development work.

SHIP Confidently

Centrally apply safety guardrails

Arch centralizes guardrails to prevent jailbreak attempts and ensure safe user interactions without writing a single line of code. You can also define and configure custom guardrails to keep users engaged on topics or tone relevant to your application requirements.

Build with Choice

Rapidly incorporate different LLMs

Build and experiment with multiple LLMs or model versions with a single and consistent interface. Arch centralizes access controls, offers high-throughput and resiliency for traffic to 100+ LLMs -- all without you having to write a single line of code.

Standards-based Telemetry

Rich tracing, metrics and logs

Arch acts as a source for several monitoring metrics related to prompts and LLMs. These metrics are compatible with OpenTelemetry destinations like Signoz, Honeycomb.io, Jaeger to help you understand all critical aspects of your AI application.

Simple to Sophisticated

From RAG to Agentic systems

Arch is designed to help you move beyond basic LLM interactions into sophisticated scenarios like building multi-turn conversations and improving the speed and accuracy of your RAG scenarios.

Use Cases

What you can build with Arch

Arch for Information Retrieval

Transform prompts into precise queries

Arch extracts intent and key details from user prompts to help you accurately retrieve relevant information from databases, APIs, and knowledge sources.

Context-Aware Processing – Extracts structured data from requests
Prompt Routing – Directs prompts to specific agents or functions for improved task performance
Learn More

Arch for agentic tasks

Easily handle agentic (action) scenarios

Arch translates free-form user input into backend function calls for executing transactional operations like placing an order, processing payments, or updating user profiles. Arch applies guardrails such as clarifying the ask, and preventing unauthorized actions

Guardrails – Prevents unauthorized or unintended operations.
Observability – Logs interactions for compliance and auditability.
Function Calls – Converts unstructured prompts into API calls.
Learn More

Arch for Agentic Routing

Detect and route prompts for accuracy

Routine tasks—such as setting reminders, scheduling meetings, and sending emails—often require multiple agent interactions. Arch simplifies this by extracting structured data from prompts and ensuring that tasks are executed by the right backend agent or API

Natural Language Task Understanding – Converts user instructions into API-based actions.
Observability & Logging – Tracks automation triggers for auditability.
Learn More

Arch for connecting with SaaS APis

Build a unified conversational interface

Arch enables users to interact with multiple SaaS applications through a unified chat interface, intelligently routing requests to the right systems while maintaining session continuity.

3rd-party SaaS Routing – Detects which service (e.g., Slack, Jira, Notion, Salesforce) should handle the request.
Secure API Orchestration – Ensures only authorized actions are executed.
Learn More
Under the Hood

Go from basic scenarios to powerful agentic apps in minutes

Go From Basic To Sophisticated

Arch is an intelligent (edge and LLM) proxy server designed for agents - to help you focus on core business objectives. Arch handles critical but the pesky tasks related to the handling and processing of prompts, which includes detecting and rejecting jailbreak attempts, intelligent task routing for improved accuracy, mapping user requests into "backend" functions, and managing the observability of prompts and LLM in a centralized way.

Get Started

Freedom to use, extend and deploy

No lock-in. No black boxes. Just an open, intelligent (edge and LLM) proxy for building smarter, agentic AI applications. Created by contributors to Envoy Proxy, Arch brings enterprise-grade reliability to prompt orchestration, while giving you the flexibility to shape, extend, and integrate it into your AI workflows.

Get Started

Proven Technology

Arch takes a dependency on Envoy and is a self-contained process designed to run alongside your application servers. Arch extend's Envoy's HTTP connection management subsystem, filtering, and telemetry capabilities exclusively for prompts and LLMs.

Proven success with companies like Airbnb, Dropbox, Google, and others.
Quick deployment and transparent upgrades.
Works with any application language such as Python, Java, C++, Go, PHP, etc.
Get Started

Task-optimized, efficient LLMs

Engineered with purpose-built LLMs to handle fast, cost-effective, and accurate prompt-based tasks like function calling, and parameter extraction to build more task-accurate agentic applications.

Function Calling is a powerful feature in Arch that allows your application to dynamically execute backend functions or services based on user prompts. This enables seamless integration between natural language interactions and backend operations, turning user inputs into actionable results.

Get Started

Task routing via Prompt Targets

Prompt Targets are a core concept Arch, enabling developers to define how different types of user prompts should get processed and routed.

By defining prompt targets, you can separate business logic from the complexities of processing and handling of prompts so that you focus on improving the quality of your application and have a cleaner separation of concerns in your code base.

Get Started
INTRODUCTION

Go From Basic To Sophisticated

Arch is an intelligent (edge and LLM) proxy for agents - to help you protect, observe, and build agentic apps by simply integrating (existing) API.

Arch handles critical but pesky tasks related to the handling and processing of prompts. This includes detecting and rejecting jailbreak attempts, intelligent task routing for improved accuracy, map user queries to "backend" APIs, and unifies the observability of prompts and LLMs in a centralized way.

OPEN SOURCE

Freedom to use, extend and deploy

No lock-in. No black boxes. Just an open, intelligent (edge and LLM) proxy for building smarter, more agentic AI applications. Created by contributors to Envoy Proxy, Arch brings enterprise-grade reliability to agentic applications, while giving you the flexibility to shape, extend, and integrate it into your AI workflows.

Get Started
BUILT ON ENVOY

Proven Technology

Arch takes a dependency on Envoy and is a self-contained process designed to run alongside your application servers. Arch extend's Envoy's HTTP connection management subsystem, filtering, and telemetry capabilities exclusively for prompts and LLMs.

Proven success with companies like Airbnb, Dropbox, Google, etc
Works with any language such as Python, Java, C++, Go, PHP, etc.
Quick deployment and transparent upgrades.
Get Started
PURPOSE-BUILT LLMS

Task-optimized, efficient LLMs

Engineered with purpose-built LLMs to handle fast, cost-effective, and accurate prompt-based tasks like function calling, and parameter extraction from prompts to build more task-accurate agentic applications. Function Calling is a powerful feature in Arch that allows your application to dynamically execute backend functions or services based on user prompts. This enables seamless integration between natural language interactions and backend operations, turning user inputs into actionable results.

Get Started
TASK ROUTING

Task-based prompt routing

Prompt Targets are a fundamental component of Arch, enabling developers to define how different types of user prompts are processed and routed within their generative AI apps. By defining prompt targets, you can separate business logic from the complexities of processing and handling of prompts so that you focus on improving the quality of your application and have a cleaner separation of concerns in your code base.

Get Started
How It Works

How It Works

Arch offers a delightful developer experience with a simple configuration file that describes the types of prompts your agentic app supports, a set of APIs that need to be plugged in for agentic scenarios (including retrieval queries) and your choice of LLMs.

Arch Gateway

Protect, observe and build agentic tasks in minutes

Define Arch Config
Define Agentic Tasks

Configure LLMs providers, guardrails and prompt scenarios that you would like to build

Deploy via Docker
Deploy Local or VPC

A docker image that you can deploy Arch in any environment: AWS, on premises, or locally

Ship an Agentic App

Ship enterprise-grade agentic apps that works at ANY scale