Getting started
What is PromptOps
PromptOps is a platform for managing your AI prompts outside of your code. When you edit a prompt on the web, it goes live in production instantly, with no code changes and no redeploy.
Why PromptOps?
- PMs can edit and test prompts themselves
- Changes reach production in under 5 minutes without an engineering deploy
- Versioning, rollback, A/B comparison, and cost monitoring are all built in
- Supports major LLMs including OpenAI, Anthropic, and Google AI
Getting started
Sign up
PromptOps is free to start. Sign up with your email or a Google account.
How to sign up
- Go to the sign-up page
- Sign up with a Google account or email and password
- Set your workspace name
- Follow the onboarding guide to create your first prompt
Getting started
Create your first prompt
After signing up you're taken to the onboarding page automatically. Pick one of five roles and the system prompt is filled in for you.
Available roles (presets)
- Customer support bot
- Content generator
- Translation specialist
- Data analyst
- Custom (write your own)
Getting started
Invite team members
On Settings → Members, enter an email and assign a role (Editor / Viewer) to send an invitation. Recipients sign up and join the workspace in a single click.
SDK integration
Auto-Init (one import line)
With Auto-Init, simply importing the SDK automatically wraps the OpenAI / Anthropic / Google AI SDKs. Your prompts are swapped dynamically for the latest version from PromptOps, with no changes to your existing code.
import '@promptops/sdk/auto';
// Your existing OpenAI / Anthropic / Google AI code works as-is
SDK integration
Run API (HTTP call)
In environments where the SDK is hard to use, you can call it directly over HTTP. Send variables as JSON to the POST /api/v1/run/{slug} endpoint and the prompt runs, streaming back the result text.
SDK integration
CLI scan
Automatically extract prompts from your existing codebase and import them into PromptOps.
npx @promptops/cli scan --queue --api-key $POP_KEY
SDK integration
Environment variables
The SDK needs exactly one environment variable.
PROMPTOPS_API_KEY=pop_...
Guides
Versioning & rollback
Every time you edit a prompt, a new version is created automatically. If something goes wrong, you can roll back to a previous version instantly.
Label system
- prod: the version live in production
- staging: the version for testing
- dev: the version under development
Guides
A/B comparison testing
In the playground you can run two prompt versions side by side and compare the results. You can also compare across different models (GPT vs Claude).
Guides
Quality evaluation (Eval)
Build a dataset and evaluate prompt quality automatically. Korean-specific criteria such as tone, particles, and loanword spelling are supported as well.
Guides
Cost monitoring
See per-prompt cost, response speed, and usage at a glance on the dashboard. Set a monthly budget and get an alert when you hit 80%.