Résumé & JD parsing
Structured extraction with normalised dates and skills.
Features
Everything between a résumé and an offer.
Four stages, one workspace — and the complete capability list at the end, all live today.
01
Screen & score
From a PDF to a ranked, explainable shortlist.
0–100 fitment score
Evidence-anchored rubrics; totals computed in code.
Fairness by construction
Anonymised first. No inference on protected traits.
Provenance on every score
Model, prompt and rubric version, stored with the result.
Plagiarism detection
Copied résumés caught against your own pool.
Interview questions
Round-specific questions from the candidate’s evidence.
02
Schedule & interview
Your Teams or Meet. Our scheduling.
Automated booking
Offer 1–5 slots; the candidate picks, no login.
Interview panels
Up to ten interviewers; clashes refused up front.
Meetings in your workspace
Rooms minted in your Microsoft 365 or Google Workspace.
Team calendar
Week, day, month, agenda. Drag to reschedule.
Reminders & no-shows
T-24h and T-1h; no-shows flagged ten minutes in.
Transcripts & evaluation
Teams recordings transcribed and scored, no bot in the call.
03
Test & proctor
Sandboxed execution. Automated proctoring.
Languages
Python, JavaScript, Java, C, C++, Go, Dart and SQL.
1,500+ question bank
11 role families, balanced by difficulty.
Isolated executors
Code never runs in-process.
Identity & presence
A second person, an identity change, a phone.
Audio fusion
Candidate speech separated from external voices.
Strike policy you control
Five warnings by default; one incident, one strike.
04
Run the pipeline
The busywork, automated.
WhatsApp + email
Booking, reminders and updates at every step.
Task manager
Tasks created by events, closed when the work is done.
Live sync
Every open tab updates the moment something changes.
Joining-risk predictor
Six-factor risk per offer.
MIS reports
Funnel, throughput, stalled candidates — live.
Security & tenancy
Per-organisation data, encryption at rest, SSO.
Complete capability list
Everything, written down.
52 capabilities across 8 areas, all live today.
Résumé & JD intelligence7 capabilities
- PDF résumé parsing into name, contact, education, experience, skills and certifications
- Normalised YYYY-MM date ranges and canonical skill names (React.js → React, k8s → Kubernetes)
- Employment gaps over six months surfaced as facts, never as judgements
- Parse quality reported as full / partial / poor, with the issues listed
- JD parsing into must-haves, nice-to-haves, seniority and department ladder
- Prompt-injection sanitisation on every AI input
- Résumé plagiarism detection against your own candidate pool
Fitment scoring & fairness7 capabilities
- 0–100 fitment score from evidence-anchored 1–5 rubrics; totals computed in code, not by the model
- Per-role threshold (70% default) that gates booking and drives automatic tasks
- Mandatory anonymisation before scoring — protected attributes stripped in code
- No inference on emotion, accent, caste, religion, university tier or gaps
- Every score stamped with model, prompt version, rubric version and threshold in force
- Enhanced (v4) and legacy prompt stacks, switchable per workspace for side-by-side comparison
- Round-specific technical and behavioural interview questions generated from the evidence
Scheduling & calendar8 capabilities
- Booking links with 1–5 offered slots, valid 7 days; candidate picks one in their timezone
- Interview panels of up to ten; double-booking refused before a slot is offered
- Live availability strip per interviewer while you pick times
- Meetings created in your Microsoft Teams or Google Meet
- Team calendar: week, day, month, agenda; drag to reschedule; the candidate is told automatically
- Workspace working hours (10:00–20:00 default) shade the grid without blocking a booking
- Reminders at T-24h and T-1h on email and WhatsApp; no-shows flagged ten minutes in
- Interview status derived from outcomes: cleared, not cleared, did not attend
Recording, transcripts & evaluation6 capabilities
- Teams recordings fetched through Microsoft Graph after the call — no bot joins
- Audio extracted and transcribed in 30 languages; code-switching, like English–Hindi, treated as neutral
- AI evaluation against the round’s rubric; English and final scores computed in code
- No confidence or emotion metrics — consistency measures statements, never the person
- Manual transcript paste and audio upload for rounds held elsewhere
- Scorecards per round on the candidate profile, with the interviewer’s notes alongside
Coding tests & proctoring9 capabilities
- Eight languages: Python, JavaScript, Java, C, C++, Go, Dart, SQL
- ~1,550-question bank across 11 role families, matched to the JD title
- Difficulty mix set by admins; hidden test cases; whitespace-tolerant output comparison
- Sandboxed executors with automatic failover; self-hosted runner for private, unlimited runs
- Browser proctoring: identity, presence, second person, phone, attention, audio fusion
- 31 signals; 5-strike default with shorter limits for phone, second person and identity change
- Evidence clips and screenshots attached to strikes; a 15-second debounce so one incident is one strike
- Platform defaults plus per-workspace overrides, applied without a deploy
- A test link survives a lost tab or a changed device; nothing rejects anybody automatically
Candidate communication4 capabilities
- WhatsApp messages on Meta-approved templates: booking, confirmation, reminders, reschedules, cancellations and coding tests
- Delivery receipts from signed webhooks; email as the fallback when WhatsApp fails
- One 08:00 digest for back-office notifications; time-critical reminders are never batched
- Calendar invites (.ics) that update in place when an interview moves
Pipeline, tasks & reporting6 capabilities
- Custom pipeline stages; only stages marked as rounds can be booked
- Tasks created by events (score cleared, link sent, no-show) and closed when the work is done
- Live sync across every open tab via server-sent events
- Joining-risk predictor: six-factor candidate form with a per-workspace form builder
- Team progress report, conversion funnel, recruiter throughput, stalled-candidate alerts
- Analytics and AI-spend visibility for admins; CSV/XLSX export from every table
Platform, security & integrations5 capabilities
- Every record scoped to your organisation; settings encrypted at rest
- Sign in with Google or Microsoft (SSO) or email + password; Clerk-backed with a safe fallback
- Operator console with TOTP MFA, IP allowlist, disposable-email blocking, password policy, audit log
- Integrations: Microsoft 365 / Teams, Google Workspace / Meet, WhatsApp Business (Telnyx), Resend, Deepgram, OpenRouter
- Per-workspace branding on candidate-facing pages