By Cesar Miguelañez —

Humanloop Shutdown: Migrate Your Prompts to Latitude

Humanloop will officially shut down on September 8, 2025. If you currently rely on it for prompt management, AI evaluations, and LLM observability, now is the time to transition.

Latitude is the best Humanloop alternative, offering everything Humanloop did plus agent orchestration, 2,500+ integrations, and improved observability in Latitude 2.0.

This guide walks you through how to migrate your prompts, agents, and workflows from Humanloop to Latitude quickly and without disruption.

What is Latitude?

Latitude is an open-source prompt engineering and AI agents platform that helps teams design, test, and deploy LLM applications faster and with greater accuracy.

It offers:

  • Prompt Management: Organize, version, and optimize all your prompts in one place.
  • Evaluations: Choose from dozens of built-in evals, with options for human-in-the-loop or LLM-as-a-judge.
  • Observability: Monitor prompt performance, latency, costs, and outcomes in real time.
  • Experiments: Run controlled tests across different models and providers.
  • Synthetic Dataset Generation: Automatically create large-scale datasets to stress-test prompts against thousands of variable inputs.

With Latitude, you get the flexibility of open source, the reliability of enterprise tooling, and the speed to go from prototype to production without friction.

Humanloop → Latitude: Quick Feature Mapping

Humanloop Feature

Latitude Equivalent

Notes

Prompt Files & Version Control

Prompt Management

Advanced versioning with branching, multiple environments (dev/staging/prod)

Datasets

Datasets Management and Synthetic generation

Import CSV or generate synthetic datasets

Evaluators

Evaluations

Dozens of built-in evals, batch evals, programmatic evals, human-in-the-loop, LLM-as-a-judge

Logs

Observability Dashboard

Track every action, including completions with full cost, token, and latency data; full transparency into all requests sent to AI providers; view sub-agent calls.

Monitoring Dashboard

Observability Dashboard

Real-time monitoring, alerts, and OpenTelemetry support

Flows / Agents

Agents & Meta-Agent Orchestration

Build agents or let Late (meta-agent) coordinate them automatically

API / SDK

API / SDK

Python & JavaScript libraries, CLI, REST API

Model Proxy

Model Integration Layer

Supports OpenAI, Claude, Gemini, Mistral, Bedrock, and more

Online Evaluations

Live evaluations

Run tests across different models and providers

Pricing: Best Open-Source Alternative to Humanloop

Humanloop used a custom, enterprise-focused pricing model. Latitude is open source, so you can self-host at no cost. If you prefer our managed cloud, you can start for free or choose a paid plan.

Quick Migration Guide: Humanloop → Latitude

Step 1: Export Your Humanloop Data (Before Sept 8!)

  • Prompts and all versions
  • Evaluation datasets and results

Logs and feedback Use Humanloop’s API to export these assets before the shutdown date.

Step 2: Set Up Latitude

Option A – Use Latitude Cloud

  • Sign up at latitude.so
  • Create your first project from the dashboard

Option B – Self-host (Open Source)

# Clone the repository
git clone https://github.com/latitude-dev/latitude-llm.git
cd latitude-llm

# Install dependencies
pnpm install

# Build shared packages
pnpm build --filter='./packages/**'

# (Optional) Copy and edit env variables
cp .env.example .env

# Start services
docker compose up -d

# Run database migrations
cd packages/core
pnpm db:migrate
cd ../..

# Start the web app (if not using Tmuxinator)
pnpm dev --filter web

Access the UI at http://localhost:3000, sign up, and confirm your email via the local MailHog at http://localhost:8025.

Step 3: Import Your Prompts

  • Create prompts in Latitude’s Prompt Management
  • Copy over templates and all versions from Humanloop
  • Add metadata and comments for context

Step 4: Recreate Evaluations

  • Upload test datasets (CSV) into Latitude
  • Rebuild evaluators using built-in types (human-in-the-loop, LLM-as-a-judge, custom metrics)
  • Run a baseline evaluation to ensure parity with Humanloop results

Step 5: Update Your Code
Before (Humanloop):

from humanloop import Humanloop

hl = Humanloop(api_key="...")

After (Latitude Cloud API):

import latitude

latitude.api_key = "YOUR_API_KEY"

Update your prompt calls to use Latitude’s API/SDK or call agents directly from the UI.

Latitude’s Advantages Over Humanloop

Better Prompt Management
Latitude already goes beyond basic prompt management with a no-code editor, AI-assisted prompt creation, and the ability to refine prompts automatically based on evaluation results.

Meta-Agent Orchestration
With the upcoming meta-agent, Latte, you’ll be able to simply describe what you need and have a complete agent built for you, ready to integrate with over 2,800 tools.

Improved Observability
Latitude’s observability is on another level: track every log, every request sent to providers, and every action your agents take, all in a fast, visual interface that makes debugging and optimization effortless.

If your team needs structured agent testing, smooth collaboration with domain experts, and deployment options that fit both fast-moving startups and regulated industries, Latitude offers a complete platform to take you from prototype to production with confidence.

Try it now for free or check our docs.