Skip to main content

Observe your AI agents with E1

E1 is an observability platform for AI agents and LLM-powered applications. With the Ellipsis SDK, you can instrument supported agent frameworks, collect execution telemetry and inspect the resulting traces in your E1 workspace. Whether you are building a single tool-calling agent or a multi-agent application, E1 gives you visibility into the operations that make up an execution.

Quickstart

Install the SDK, configure your agent and export your first trace.

TypeScript SDK

Explore SDK requirements, configuration and supported integrations.

First example

Follow a working LangChain agent example powered by Google Gemini.

What can you observe?

The Ellipsis SDK collects telemetry from supported framework executions and sends it to the E1 platform.
  • Agent and chain executions — follow the operations performed during an agent run.
  • LLM invocations — inspect model and provider information, invocation duration and token usage when reported by the framework.
  • Tool calls — see tool operations associated with an agent execution.
  • Trace and span hierarchy — understand how operations relate to one another through parent-child spans.
  • Performance and usage — investigate execution duration, latency, token consumption and prices per run.
  • Failures — inspect errors and failed executions.
The SDK batches telemetry for export and includes retry mechanisms for transient network failures.

How E1 fits into your application

The integration has three main parts:
  1. Your agent application runs using a supported framework and an LLM provider.
  2. The Ellipsis SDK instruments supported framework callbacks, organizes execution events into traces and spans, and exports telemetry.
  3. The E1 platform stores and presents the collected telemetry in your workspace.
The SDK is designed to add observability without requiring you to rewrite your agent workflow.

Current integration scope

The current TypeScript SDK supports LangChain and LangGraph instrumentation through LangChain callbacks. It can capture telemetry for different LLM models supported by LangChain, including models from OpenAI, Google Gemini, Anthropic and OpenRouter.
Provider-specific instrumentation outside the LangChain framework is not currently supported. The SDK currently requires Node.js 22 or later.

Start integrating

The fastest way to see E1 in action is to connect an agent to your workspace.
1

Create an E1 agent

Sign in to E1, open your organization’s workspace, create an agent and generate its API key.
2

Install and configure the SDK

Follow the Quickstart to install @ellipsis1/sdk and configure the ingestion URL and agent API key.
3

Instrument your framework

Create the LangChain handler with ellipsis.langchain() and pass it to your agent invocation as a callback.
4

Inspect your telemetry

Run your application and open the corresponding agent in E1 to inspect the telemetry exported by the SDK.

Next steps