We don't algorithmically rank humans.
Knockhaus is the operating system used by your hiring manager to organize applications. We deliberately do not use AI to score, rank, or recommend candidates. Every hiring decision is made by a human.
No algorithmic ranking. No fit scores. No AI recommendations.
Some recruiting platforms compute a numeric "fit score" or "match percentage" for each applicant. We don't. We don't auto-rank candidates. We don't hide low-scoring applicants from the manager's view. We don't auto-reject below a threshold. We don't use AI to evaluate video interviews.
These are deliberate design constraints in the platform, enforced at the database level. The recruiting tables physically cannot store a candidate score — score, fit, match, rating, and recommended are not legal column names in our schema. Every stage transition (applied → screened → interviewed → offered → invited) is logged with the user ID of the manager who moved it. If a candidate ever progresses without a human signing off, our system flags it as a violation.
Surface observable facts. The hiring manager decides.
When an applicant submits an application, AI may extract observable facts from their cover letter and résumé. These appear in a section called Quick facts on the manager's applicant view. Examples:
Prior employer: SunrunYears of D2D experience: 3Languages: English, SpanishMentions of "commission" in cover letter: 4Cover letter length: 230 words
These are observable — either restated from what the candidate themselves wrote, or counted directly. They are not evaluations. They are the same information a human reviewer would extract by reading the application, just surfaced faster.
Three non-negotiable rules for our engineering team.
- Show ingredients, never recipes. Surface observable facts. Never compute a composite score, fit percentage, or single number summarizing a human.
- Sort by time, never by quality. The default sort in every recruiting view is chronological. Managers can apply their own filters; the system never ranks for them.
- The human holds the pen.Every stage transition is logged as a manager action. If our system ever moves a candidate without a real user ID, it's a violation and the move is rejected.
Request human-only review at any time.
Every public application form includes a "Request human-only review" checkbox. Because we already don't use AI to score candidates, this toggle is effectively a no-op — but we record it permanently in the consent log so you have proof that you exercised the option, in case AI use in hiring becomes more common in the future.
This satisfies the "alternative selection process" requirement of NYC Local Law 144 and the alt-path provisions of similar laws in Colorado, Illinois, and the EU.
The regulatory weather is changing.
Several jurisdictions now regulate "Automated Employment Decision Tools" (AEDTs) — algorithmic systems that score, rank, or filter job applicants. The most prescriptive is NYC Local Law 144 (active since July 2023), which requires annual independent bias audits, public audit summaries, candidate notice, and an alternative-review path. EEOC Title VII applies federally. Colorado SB 24-205 takes effect February 2026. The EU AI Act classifies employment AI as "high-risk" for August 2026.
By not using AI to score humans in the first place, we avoid the underlying problem these laws regulate. Our hiring managers see better data; candidates get reviewed by a real person; nobody's career is decided by a number a machine made up.
If you applied and want details about how we handled your application.
Email privacy@knockhaus.app. We'll show you what data we stored, when it was reviewed, and by whom. If you applied at a specific organization, you can also request this directly from that organization — they own the application data; we host it on their behalf.