How teams ship
reliable AI products

Laminar is a unified open-source platform for tracing, evaluating, and labeling LLM products.

Backed by

Teams that ship better LLM products with Laminar

I can attest to it being the only reliable and performant LLM monitoring platform I've tried. Founding team is great to talk to and super responsive.

Tommy He

CTO, Clarum

Laminar's evals help us maintain high accuracy while moving fast, and their team is incredibly responsive. We now use them for every LLM based feature we build.

Hashim Rehman

CTO, Remo

Laminar's tracing is genuinely great. So much better than the others I've tried.

Michael Ettlinger

CTO, Saturn

Trace

Tracing your LLM application provides visibility into every execution step while collecting valuable data for evaluations, few-shot examples, and fine-tuning. With Laminar, you can start tracing with just 2 lines of code.

import { Laminar, observe } from '@lmnr-ai/lmnr';

// automatically traces common LLM frameworks and SDKs
Laminar.initialize({ projectApiKey: "..." });

// you can also manually trace any function
const myFunction = observe({name: 'myFunc'}, async () => {
...
})

Effortless Observability

Add 2 lines of code to trace all LLM calls and traces. Traces are sent in the background via gRPC with minimal performance and latency overhead.

Start tracing

Online evaluations

Setup LLM or Python online evaluators to process each received span. Evaluators automatically label spans, which is more scalable than human labeling.

Setup online evaluations

Dynamic few-shot examples to improve prompts

Build datasets from traces for evaluations, fine-tuning and prompt engineering. Enhance prompts by retrieving semantically similar examples from indexed datasets.

Create a dataset

Serverless LLM pipelines

Our pipeline builder is an incredible prototyping tool. It lets you quickly build and iterate on both simple prompts and complex LLM chains. After that

Deploy LLM pipeline

Fully open-source

Laminar is fully open-source and easy to self-host. Get started with just a few commands.

Self-host Laminar