Comment by navaed01

“ For reference, our team at RunLLM has spent roughly 70% of our engineering cycles in the last quarter on data engineering — everything from pre-processing data at ingestion time to implementing hybrid search to reranking results.”

This to me is why I am actually bullish on building ‘LLM wrappers’ that are sustainable businesses. While you might not be developing new technology, you can be adding value to a client who can use chat gpt and “talk to your docs” but not the surrounding work to make it frictionless for their workflow or company and actually build a business. It’s entirely possible that AI will get so good that it makes it easy to all the surrounding tasks too.