Hi everyone. I've been exploring RAG pipelines for a small project and have been prototyping with LlamaIndex for the last few weeks. It's been great for getting something working quickly.
I'm now trying to decide whether to commit to it long-term. My main concern is cost scaling. For those with experience, how does the long-term cost of using LlamaIndex's paid features (like their data connectors or hosted services) compare to the engineering effort of building and maintaining a custom pipeline with open-source tools? I'm comfortable with basic Docker and Python, but building a robust system from scratch feels like a big leap.
I’d love to hear about real experiences over a 6-month period. Was the initial time savings worth the potential ongoing costs? Or did rolling your own become more cost-effective after the initial hump? Thanks for any insights you can share!