Just finished a PromptLayer integration for a B2B client's Django app. The marketing says it's "seamless." It's not. But it's worth the setup time for the monitoring alone.
Here's the real steps, skipping the fluff. First, ditch the pip install 'promptlayer' alone. You need the openai package too. Configure your settings.py with the PromptLayer API key and set it as the default openai client. Then, wrap your existing openai calls. The key is to pass `pl_tags` with your Django model instance IDs or user IDs. That's how you get useful tracking later. Their dashboard will then show cost, latency, and token usage per customer or feature. The webhook for logging requests is solid, but test it with a staging environment first—their docs overpromise on the setup speed.