Claude Code recently integrated native support for the Language Server Protocol (LSP), a significant development for AI coding tools. LSP allows an integrated development environment (IDE) to understand code more intelligently. Features like "go to definition" or "find all references" are now accessible to Claude Code, enhancing its capabilities beyond simple text manipulation. Previously, AI coding tools relied on parsing code as text, which limited their understanding. With LSP, Claude can leverage the same code intelligence that helps human developers navigate complex codebases. This shift poses a threat to many startups that have emerged to improve AI's code comprehension. Tools that wrapped LSP or created semantic code search systems are now at risk. The author, who previously worked on a project called Sagrada meant to enhance AI's code understanding through knowledge graphs, found that developments in Claude Code quickly overshadowed his efforts. As base models like Claude continuously improve, any AI tool that claims to enhance understanding may find its value proposition weakened. Several categories of tools should be concerned. AI IDEs like Cline and Cursor, which differentiate themselves by providing better context to AI, may struggle as LSP support makes that context less critical. Their survival will depend on whether their strengths lie in unique integrations rather than just code comprehension. Frameworks like LlamaIndex and LangChain may fare better, as they offer broader applications beyond just improving AI’s understanding. The landscape is shifting rapidly, and tools focused solely on enhancing AI comprehension are likely to find themselves outpaced by the evolving capabilities of base models.