-
Need a tool/advice to bulk categorize thousands of keywords (LLMs keep skipping data)
Posted by on January 25, 2026 at 10:59 amI’m trying to sort a massive list of keywords into specific groups based on my own rules.
I’ve thrown this at GPT, Gemini, NotebookLM, and even Deep Research (using Excel/PDF inputs). The issue isn't the logic, it's that the models get lazy and only process a fraction of the list before stopping or hallucinating.
Does anyone know how to make these tools process 100% of the file? Or is there a better non-LLM tool for this kind of bulk sorting?
Appreciate the help!
-
1 Reply
-
Have you considered chunking these (of course depending on how technical you are)?
Also just saw the Claude Code can interact with Excel now – maybe that’s one way?
-
this is less an intelligence problem and more a batching and determinism problem. most general llm interfaces are optimized for “good enough” answers, not guaranteed full coverage, so they will always try to summarize or skip when the list is large. for thousands of keywords, you are usually better off chunking deterministically and forcing strict input output structure, or stepping outside chat tools entirely. people have decent results with simple rule based scripts, regex, or spreadsheet logic first, then using an llm only for the fuzzy edge cases. if the rules are truly yours and explicit, code beats language models every time for this kind of task. the llm works best as a reviewer, not the sorter.
Log in to reply.