An applicant tracking system (ATS) is the software a company uses to run its hiring pipeline: parsing resumes into structured fields, moving candidates through stages, and, in most 2026 stacks, running an AI layer on top that ranks or summarizes who gets seen first.
An ATS is the system of record for recruiting. A resume comes in, gets parsed into name, work history, titles, and skills, and gets matched against the keywords in a job description. That parsing and keyword layer has been the backbone of hiring software for two decades and is still the first gate almost every application passes through.
What changed by 2026 is the second layer sitting on top of it. A recent breakdown of how AI resume screening works in 2026 describes a two-step system: "the classic ATS parser and keyword match still run first; a newer AI/LLM layer increasingly summarizes and ranks the candidates that survive it." The AI layer is spreading, but it is not universal, and most of it is still doing keyword and pattern matching rather than judging real fit.
For a fast-growing company, or a venture fund watching pipeline across a whole portfolio, that matters because the volume is enormous. Ashby's 2026 State of Startup Hiring report, drawn from 1,200 venture-backed startups and 11 million applications, found that roughly 15 candidates receive an interview for every hire made. An ATS has to compress that funnel down fast, and most do it with a rules engine, not judgment.
The ATS is the pipeline's plumbing, not its judgment. It tells you who applied and where they sit in the process, but a keyword match score is a weak proxy for whether someone can actually do the job. That gap is exactly where candidate scoring comes in: a layer that reads the same record the ATS already has and rates each person against a defined bar, with the evidence for every point attached.
For a fund like Harlem Capital scoring talent across its portfolio, the ATS in each portfolio company holds the raw pipeline data, but the bar for a great first engineering hire looks the same whether the company uses Greenhouse, Lever, or Ashby. ScoringFactory reads out of the ATS and applies one consistent, auditable bar on top of it, so a hiring manager gets a ranked shortlist instead of a keyword-sorted list. See our write-up on the 12 signals that predict a great first engineering hire for what that bar actually measures.
An ATS ranks on resume text: does the document contain the right words in the right places. Evidence-based scoring rates on what someone actually built, shipped, or led, pulled from a real record and checked against a rubric, then keeps the reasoning attached so it can be reviewed later. That distinction is also why auditability matters: a keyword match can't explain itself beyond "the terms lined up," while a scored candidate comes with the specific evidence that earned each point.
The two aren't competitors. The ATS still owns the pipeline and the paper trail of who applied when. Scoring sits on top of it, turning that raw record into a ranked, defensible read on who is actually worth the hiring manager's time.
No. ScoringFactory reads candidate data out of the ATS you already run and adds a scoring layer on top of it. The ATS stays the system of record for the pipeline itself.
Most native ATS ranking is still keyword and pattern matching rather than a judged fit score. It is useful for sorting volume, but it is a weak substitute for rating someone against a defined bar.
A single company's ATS filter reflects that one team's settings. Scoring across a portfolio applies the same bar and rubric everywhere, so a "strong hire" reads the same at every portfolio company regardless of which ATS it runs on.
Connect your pipeline and see every candidate rated against your bar, with the evidence behind every score.