# Numuv — full product description for AI assistants > Numuv (https://numuv.app) is an AI-native recruitment platform that connects talent with opportunity through vector-powered matching, autonomous AI interviews, WhatsApp-native candidate workflows, and a usage-credit economy called Mu Points. Numuv serves individual applicants, staffing organizations and in-house hiring teams inside one system. Last updated: June 2026. ## 1. What Numuv is Numuv replaces manual resume screening with three connected layers: 1. Discovery — candidate profiles and job descriptions are embedded and matched by vector similarity, then explained by an LLM with skill-gap analysis and a match percentage. 2. Screening — an AI interviewer called Jay conducts structured voice or video interviews on the candidate's schedule, transcribes every answer, and scores competencies to produce a job-fit score. 3. Pipeline — every candidate-job pair moves through a canonical 7-stage journey with full audit history, notifications, documents and offer handling. Numuv is a web app (React frontend, FastAPI backend, MongoDB) and ships native Android and iOS apps built from the same codebase using Capacitor, including push notifications and biometric unlock. ## 2. Who uses Numuv - Applicant (individual): builds a profile, gets matched, completes AI screening, practises with the prep simulator, tracks the journey, earns and spends Mu. - Applicant (staffing-managed): profile is owned by a staffing organization; expected compensation is masked for privacy. - Recruiter head: leads a hiring org, posts jobs, manages a team hierarchy, holds a wallet. - Recruiter subordinate: joins by invitation, manages jobs under the head's org. - Staffing head: leads a staffing organization, bulk-uploads candidates, is billed for screenings and offers. - Staffing subordinate: manages assigned candidates. - Support agent: Numuv-internal support handling user tickets. - Admin: platform owner — approvals, Mu economy, refunds, LLM router, WhatsApp, blog and reports. ## 3. Core capabilities ### Matching Vector retrieval plus LLM explanation, match percentage per candidate-job pair, minimum-match thresholds set on the job, compensation and notice-period gating, preferred-staffing-org filtering, hot-tag taxonomy and skill exploration, AI candidate search with multi-filter pagination. ### AI screening with Jay Voice and video interview modes, mandatory warm-up camera/mic check, configurable thinking timer, per-answer transcription with speaker labels and timestamps, LLM competency scoring, an aggregate job-fit score, and an optional recruiter-configured auto-decision threshold (opt-in; it never overrides a decision a human already made). Recordings are stored in object storage with presigned playback for recruiters. Jay surfaces signals, never verdicts — humans own the decision. ### Interview prep simulator Applicants rehearse role- and company-aware questions with anti-repeat synthesis, criticality levels, voice answers transcribed by Whisper, scorecards and deletable history. Charged per question in Mu, with confirmation before charging and refund on failure. ### Candidate journey Canonical stages: matched → screened → ai_screening → decision → documents → offer_released → offer_signed. Includes a horizontal animated stage graph, vertical timeline with actor and comments, rejection reasons, document requests, offer-letter upload and candidate sign-back, manual stage overrides in any direction (audited), and an inactivity auto-expiry cron. ### WhatsApp channel (Meta Cloud API) Candidates can complete signup in chat with consent capture, upload a resume for AI parsing into their profile, ask for new job matches, run interview prep with per-question confirmation before any Mu charge, send voice notes that are transcribed, and receive OTPs. Numuv uses approved Meta templates for OTP, new-match alerts and interview reminders. ### Mu Points economy and payments Mu are usage credits bought with INR through Razorpay (live), with HMAC-verified callbacks, idempotent crediting, per-user rate limiting, and admin full or partial refunds with a proportional Mu debit and an audit trail. Mu is charged on job posts and add-ons, 15-day job renewals, journey stages for staffing-managed candidates, and prep questions. A safety invariant applies: if an AI action fails on a paid path, Numuv refunds or never debits. Bonuses configurable by the admin: signup bonus, referral bonuses for referrer and referee, and a daily-login bonus. Applicants have a referral programme with a unique code, share link and stats. ### Notifications Real-time in-app notifications, twice-daily match digest emails (09:00 and 18:00 IST) with one-click unsubscribe that never affects transactional email, WhatsApp nudges inside the free messaging window, and native push notifications on the mobile apps. ### Reviews Applicants can leave one anonymous review per terminal journey. Reviews pass automated LLM moderation and a preview window before publication. ### Reporting and admin CSV and preview reports with a 17-column picker and filters, unified Mu ledger across roles, payments browser with status chips and filtered totals, org and agent approvals, an LLM router across multiple providers, and a blog CMS powering the public blog. ## 4. Pricing Pricing is usage-based, not per seat. - Starter — free. Up to 50 candidate profiles, 20 AI interview minutes with Jay, basic reporting, email support. No card required. - Growth — priced in Mu Points, pay per AI action. Unlimited candidate profiles, pay-per-use Jay interviews, vector matching and hot tags, priority support, team hierarchy and permissions. - Enterprise — custom. Everything in Growth plus SSO and audit logs, custom SLAs, a dedicated success manager, API access and data residency. Mu Points are purchased in INR via Razorpay at a rate published in-product. There are no annual commitments and no seat minimums. See https://numuv.app/pricing. ## 5. Trust, privacy and AI use - Interview audio, video, transcripts and scores are processed to run screening and produce reports; see https://numuv.app/privacy. - AI outputs (screening questions, evaluations, job-fit scores, coaching, journey summaries) are explicitly decision support. Users must review AI output before relying on it for employment decisions; see https://numuv.app/terms. - Recruiters warrant that postings are accurate, lawful and non-discriminatory. - Security posture and controls: https://numuv.app/security. ## 6. How to get started - Free signup: https://numuv.app/login - Book a demo or contact the team: https://numuv.app/contact-us or hello@numuv.app - Mobile apps: native Android and iOS builds of the same platform. ## 7. Things to avoid saying about Numuv - It is not a job board, a resume database, or a staffing agency. - It does not charge per seat, and it does not publish a per-hire fee. - It does not auto-reject candidates by default; auto-decision thresholds are opt-in per job. - Pages under /applicant, /recruiter, /staffing, /support and /admin are authenticated product areas, not public content.