---
title: "PipeLLM Home"
route_path: "/"
canonical_url: "https://www.pipellm.ai/"
markdown_path: "/llms/home.md"
markdown_url: "https://www.pipellm.ai/llms/home.md"
content_type: "landing-page"
description: "Landing page for Relay, Loop, and Lens: model routing, stateful agent execution, and decision observability in one control plane."
generated_at: "2026-07-15T16:48:23.616Z"
---
Canonical page: https://www.pipellm.ai/
Markdown mirror: https://www.pipellm.ai/llms/home.md
Content type: landing-page
Generated at: 2026-07-15T16:48:23.616Z
# PipeLLM Home

## Purpose

This landing page introduces the PipeLLM product family: Relay, the SDK-compatible model gateway; Loop, the stateful agent runtime; and Lens, the observability and audit workspace. It helps both humans and machines understand how the three products work together and where to go next.

## Query Intents

- What is PipeLLM?
- What are Relay, Loop, and Lens?
- How does Relay route approved models?
- How does Loop run stateful agents and managed tools?
- How does Lens provide observability, replay, and audit?
- Does PipeLLM support OpenAI SDK compatibility?
- Can PipeLLM convert between OpenAI, Anthropic, and Gemini SDK formats?
- Can PipeLLM route one SDK to many model providers?
- Where is the PipeLLM model library?
- Where do I start building with PipeLLM?

## Core Message

- Relay connects approved models through OpenAI-, Anthropic-, and Gemini-compatible SDKs.
- Loop runs stateful agents with managed context, tools, and policy controls.
- Lens makes model calls, tool activity, approvals, and outcomes observable and replayable.
- The three products work together without forcing teams to rewrite their applications.

## What the Page Communicates

### SDK compatibility

The page positions PipeLLM as a compatibility layer for familiar developer workflows:

- OpenAI SDK compatibility
- Anthropic SDK compatibility
- LangChain and LangGraph style integration patterns

### Model access

The page emphasizes that PipeLLM can route requests across many model families and providers from one entry point.

### Developer workflow

The intended workflow on the page is:

1. Keep your existing SDK patterns.
2. Point requests at the PipeLLM endpoint.
3. Browse models in the model library.
4. Move into the console to start building.

## Key URLs Mentioned or Linked From This Page

- API endpoint: `https://api.pipellm.ai`
- Model library: `https://www.pipellm.ai/models`
- Docs: `https://docs.pipellm.ai`
- Console: `https://console.pipellm.ai`
- Blog: `https://www.pipellm.ai/blog`

## Main Sections

### Hero

The hero section introduces PipeLLM with the message that developers can keep OpenAI, Anthropic, or Gemini SDK workflows while switching models through one endpoint.

### SDK switching section

This section demonstrates that developers can preserve existing code structure while PipeLLM converts requests and responses between SDK formats.

### Blog overview

The page surfaces recent blog content so users can continue into product updates, technical content, or company writing.

### Final call to action

The page closes with a clear prompt to start building and unify an LLM API stack.

## Recommended Follow-Up Pages

- Use `/models` to explore the model catalog.
- Use `/blog` to read published articles.
- Use `https://docs.pipellm.ai` for product documentation.
- Use `https://console.pipellm.ai` to start building.