I think the key quote is:<p>>"Claude's extensive context window has also transformed their approach to handling large codebases. When the <i>200K context window</i> was released, Hedley notes they "<i>ripped out the entire RAG</i> and just put it in the context window instead and it went from <i>60 percent accuracy to 98</i>."<p>RAG = Retrieval Augmented Generation<p>Related:<p>What is retrieval-augmented generation, and what does it do for generative AI?:<p><a href="https://github.blog/ai-and-ml/generative-ai/what-is-retrieval-augmented-generation-and-what-does-it-do-for-generative-ai/" rel="nofollow">https://github.blog/ai-and-ml/generative-ai/what-is-retrieva...</a><p>Retrieval Augmented Generation:<p><a href="https://en.wikipedia.org/wiki/Retrieval-augmented_generation" rel="nofollow">https://en.wikipedia.org/wiki/Retrieval-augmented_generation</a><p>Retrieval-Augmented Generation for
Knowledge-Intensive NLP Tasks (2021) (apparently the original paper where the term 'RAG' was coined):<p><a href="https://arxiv.org/pdf/2005.11401" rel="nofollow">https://arxiv.org/pdf/2005.11401</a>
A good deal of software development is clarifying the required specification. This alone takes a lot of work and a lot of coding! If you don't know the nitty-gritty of what you really need, you can't get it at 100x or even at 10x speed.
When it comes to financial accounting work, there is just no room for buggy or sloppy code. The customer will go away forever at the first instance of being billed incorrectly, also seeking a refund from the credit card.