Understanding Nuance: Same Skills, Different Candidates

Two resumes can list identical skills and describe completely different people. The difference lives in the context — and context is exactly what Mosaic is built to read.

The Skill List Is Not the Candidate

Picture two resumes that both say “Python, SQL, data analysis.” One belongs to a researcher who builds statistical models to answer open-ended questions. The other belongs to an operations engineer who automates reports and keeps pipelines running. Same three keywords. Different instincts, different strengths, different hires.

Any experienced recruiter reads that difference instantly — not from the skill list, but from everything around it: the verbs, the projects, the scale, the story. The question is whether your software can read it too.

Two Levels of Reading, at the Same Time

Mosaic reads every document the way a careful human does — at two levels at once.

The whole story. First, Mosaic forms an impression of the entire resume: the arc of the career, the kind of problems this person gravitates toward, the environments they've thrived in. It does the same for every job description — capturing not just the requirements list but the intent behind the role. A req that asks for “someone who can apply technology to business goals” is asking for a mindset, not a keyword. Mosaic reads it that way.

Every detail in context. Second, Mosaic looks at each individual skill, tool, credential, and responsibility — and keeps the surrounding sentences attached. “Python” next to “built machine-learning models for clinical trials” is a different fact than “Python” next to “wrote scripts to rename files.” The word is the same; the meaning isn't. Mosaic stores the meaning, not just the word.

Why Both Levels Matter

Either level alone falls short. A system that only reads the whole document can tell you two resumes “feel similar” but can't say why — and can't check a specific requirement. A system that only extracts skills can check requirements but misses the person behind them. Mosaic's matching uses both: the big picture to understand fit and intent, the detailed context to verify each requirement with evidence.

In practice, that means:

• The researcher and the engineer rank differently for a modeling role and an automation role — even though their skill lists match word for word.
• Depth is visible. Five years administering a platform reads differently than one bullet point mentioning it, because the context says so.
• Intent is honored. When a job calls for business judgment, candidates who demonstrate it in how they describe their work rise — even if they never use the phrase.
• Explanations stay concrete. Every match can point to the actual language in the resume that supports it, because the language was never thrown away.

Nuance Plus Structure: The Complete Picture

This is the third piece of the Mosaic approach. Entities give the system a vocabulary built from your company's own hiring (Part 1). The talent graph connects those entities into a map of relationships — who has what, who worked where (Part 2). And nuance — meaning captured at both the document level and the detail level — is what makes every point on that map trustworthy.

Together they do what no keyword engine and no generic AI score can: evaluate candidates the way your best recruiter would, at the speed and scale of software.

The Bottom Line

Keywords tell you what words a candidate used. Nuance tells you who the candidate is. Mosaic reads both — so your shortlist reflects people, not word counts.