Resources/Glossary/AI
Natural language processing (NLP) is the branch of AI that deals with human language. Resume parsing, keyword extraction, semantic search and recruiting chatbots all use it, and today most NLP runs on large language models.
In one line
The branch of AI that deals with human language. Resume parsing, keyword extraction and chatbots all use it, and today most NLP runs on large language models.
NLP covers any task where software has to make sense of words: pulling a job title and dates out of a resume, recognizing that “RN” and “registered nurse” mean the same thing, sorting candidate replies into interested or not, or answering an applicant’s question in a chat window. Each of these jobs looks simple to a person but takes real work for software to get right across thousands of different resumes.
Before large language models, NLP tools were built from rules and smaller trained models, one task at a time. They were fast but brittle, which is why older parsers mangled resumes with tables or unusual layouts. LLMs handle variety much better, so many vendors have rebuilt their NLP features on top of them. NLP still makes mistakes with abbreviations, niche skills and context. A parser can read “Java” in “Java coffee shop barista” as a programming skill if nothing checks it.
Every search, filter and match in your ATS depends on how well its NLP turned resumes into data. If the parser misses skills or dates, good candidates disappear from your searches. Understanding NLP helps you test tools properly, write job ads machines read correctly, and know when to search with Boolean strings instead of trusting a semantic match.
Searching the ATS for “Kubernetes” returns 12 results, which seems low to the recruiter. She runs the same search on the raw resume text and finds 31. The ATS parser had dropped skills listed inside a two column table. The team reports it to the vendor and, meanwhile, adds a full text search step to every technical search.
Resume parsing, candidate search, job description writing, chatbots that answer applicant questions, survey comment analysis and sorting email replies. Any task where software reads or writes words uses some form of NLP.
No. NLP is the field; a large language model is one powerful tool used within it. Most modern NLP features run on LLMs, but some older or faster tasks still use smaller models built for one job.
Usually because the resume uses tables, columns, text boxes, headers or images that the parser was not built to read. Simple, single column resumes parse best. Newer parsers built on LLMs handle odd layouts better.
Written by recruiters at Work Insiders. Part of the recruiting and HR tech glossary. Updated October 2026.