Answers to frequently asked questions

Blade is an AI agent with 17 hiring intents — it opens a role, writes and posts the job description, searches your candidate database in natural language, launches interviews, chases candidates and answers questions about your pipeline. You talk to it by voice or text in 15+ languages. It acts inside the platform rather than handing you a draft to paste somewhere else.

A chatbot answers a question and stops. An AI recruiter takes the action: it can post the role, source candidates, send the interview, score the response and move the candidate to the next stage. The practical test is whether the tool changes the state of your pipeline on its own, or only produces text you then act on.

Autopilot runs the pipeline end to end — sourcing, outreach, screening, interviewing and shortlisting — and hands a human the shortlist to approve. You set which stages run unattended and which need a person. It is switched off by default; nothing runs autonomously until you turn it on for a specific role.

Semantic search across the candidate database returns in about 2 seconds at p95 in production, measured in 2025. You search in plain language — "React developers who have shipped payment integrations and are open to Berlin" — rather than by boolean keyword strings.

Yes, and the restrictive setting is the default. Automated rejection is off out of the box: where a score falls below a threshold the platform proposes a rejection and a person confirms it. Each intent can be enabled per role, and every automated action is written to an audit trail with the reasoning attached.

Yes. Engagement chains fire steps over email, Slack, WhatsApp and outbound voice calls, and the agent can be driven from chat rather than the web app. The no-code builder means a recruiting coordinator sets these up without engineering help.

You configure 5–7 questions per role and candidates record answers on their own schedule from a single link. You review whenever it suits you, with transcription and per-competency scoring already attached. The scheduling round-trip disappears, which is where most of the calendar time in early-stage screening goes.

Any device — desktop, laptop, tablet or phone, in any modern browser, with nothing to install. The interface is mobile-first because most candidates complete on a phone, which matters in retail, hospitality, logistics and other deskless hiring.

It calls the candidate, runs a first-round screening conversation and scores the answers, responding in under 500 ms so the call does not feel like an IVR. It is used where volume makes human phone screens impractical — high-volume retail and call-centre hiring, and outbound sourcing where the first contact has to happen quickly.

Nine adaptive cognitive games covering pattern recognition, working memory, task switching, processing speed, spatial reasoning, prioritisation, ethical judgement, attention to detail and numerical reasoning. They run computerised-adaptive (CAT) across five difficulty bands and 95+ calibrated tasks, so the battery converges in about 10 minutes rather than the 60 a classical psychometric test needs.

General cognitive ability is the strongest single predictor of job performance in the industrial-psychology literature; personality inventories predict far less for most roles. Personality tests are popular because the output reads as insightful, not because it forecasts anything. We score cognition and job-relevant competencies, and leave typology out.

The answer is transcribed, scored per competency against your rubric and summarised in a paragraph. Every score carries its reasoning, so a reviewer can check the basis and override it. The score is a ranking aid for a human decision, not the decision.

Yes. Build multi-format tests — multiple choice, open response, code submission — and the platform auto-grades and ranks them with a full audit trail. There is also a question bank and role templates if you would rather start from something and edit it.

Screening applies declarative knockout rules plus multi-layer detection of AI-generated and dishonest responses, and take-home submissions get integrity checks. Detection is reported as a signal on the response for a human to weigh, not as an automatic rejection.

More than 30 connectors, including LinkedIn, GitHub, Stack Overflow, Indeed, Apollo and Hunter, with deduplication across them so one person found in four places is one record. Premium covers 15 connectors; Premier and Enterprise cover the full 30+, and Enterprise can add custom connectors.

Yes. Sourcing and the internal database both take natural-language queries instead of boolean strings, and the Talent Graph links people, skills, companies and past interactions so a search can follow relationships rather than matching keywords. There is a REST API over the same index.

That is the case it is built around. Async video and the voice agent remove the scheduling bottleneck, knockout rules cut the funnel before a human looks at it, and engagement chains handle follow-up at volume. Premier covers up to 50 open roles and 800 interviews a month; Enterprise is unmetered.

Yes. Roles, career pages and candidate flows are per-location, the agent and interviews run in 15+ languages, and data residency is chosen per customer. Multi-location and multi-brand groups run one workspace per entity under a single account rather than separate subscriptions.

About three minutes. Career pages are generated from your roles and branding with no developer involvement, on a custom domain, and stay in sync as roles open and close. Job descriptions can be generated from a short brief and edited before they go live.

Predictive hiring means checking whether your screening actually forecasts job performance, instead of assuming it does. HRBlade keeps scoring after the hire — through onboarding, 30/60/90-day surveys and performance reviews — and correlates the pre-hire score against what the manager reports later.

It is the correlation between the AI interview score and manager-verified performance at six months, measured in 2025 across a benchmark cohort of 2,400 hires. It is a benchmark on our own cohort, not a guarantee for a given customer: the model recalibrates on each customer's hiring history, so your own number is the one that matters and it takes about two quarters of outcomes to produce.

Model calibration happens per customer on that customer's hiring history, and candidate data is never used to train, fine-tune or improve any AI model, ours or a vendor's. Nothing about your candidates moves to another customer's model.

Funnel analytics, conversion reporting by stage and source, comparative and correlation analysis across roles, and raw export by CSV, JSON or the REST API. Enterprise adds quarterly business reviews with a solutions engineer against your own numbers.

Standard is $159 a month ($119 on annual billing), Premium $319 ($249 annual) and Premier $999 ($699 annual). Enterprise is quoted. Prices are shown in USD, EUR or GBP, and every plan starts with a 14-day free trial without a card.

Active job openings, user seats, interviews per month, AI credits and video or voice minutes — not headcount at your company. Standard covers 5 open roles and 1 seat; Premium 10 roles and 5 seats; Premier 50 roles and 20 seats; Enterprise is unmetered.

Yes — upgrade any time and it applies immediately, prorated. Downgrades take effect at the start of the next billing cycle, so you keep what you paid for until then.

14 days of full Premium access, no credit card. The trial is the real product with real candidates, not a sandbox, so the roles you open during it carry over if you subscribe.

Credit card — Visa, Mastercard and Amex — and invoiced bank transfer for annual contracts on the Premier and Enterprise plans.

Yes. Annual billing is roughly 25–30% below the monthly rate on every self-serve plan: $119 against $159 on Standard, $249 against $319 on Premium and $699 against $999 on Premier, quoted per month.

Data is encrypted in transit and at rest, access is role-based, and every AI scoring action is logged with the reasoning behind it for GDPR Article 22 explainability. A Data Processing Agreement is available on request.

You pick the region: EU-Frankfurt and US-East are standard, UK and UAE are available on Enterprise. Customer data stays in the chosen region.

Yes, on Enterprise — on-premise or private cloud, which is how banking, healthcare, public-sector and defence customers usually run it. The platform, models and data sit inside your perimeter.

No. There is no facial recognition, no facial expression or micro-expression analysis, no gaze tracking and no video-based scoring anywhere in the platform. CVs are converted to plain text before processing and a photograph in a CV is never sent to any model.

No. Article 5(1)(f) of the EU AI Act prohibits inferring emotions from biometric data in the workplace, and the European Commission's February 2025 guidelines confirm that recruitment is covered. No HRBlade feature infers emotional state, mood, confidence, enthusiasm or stress, and no candidate is scored on any such attribute. We publish a signed attestation to that effect.

No. There is no analysis of tone, pitch, pace, pauses, prosody or accent. Audio is converted to text by a speech-to-text engine and only the text is assessed, so a candidate is not scored on how they sound.

Not by default. Automated rejection is switched off; below a configured threshold the platform proposes a rejection and a person decides. Where an employer chooses to enable automation, the decision is logged with its reasoning and the candidate's Article 22 rights apply.

HRBlade is GDPR and CCPA compliant, and publishes an AI Transparency Statement under Regulation (EU) 2024/1689 naming the provider of the system, along with a DPIA template, a processing record, a subprocessor list and works-council documentation. Procurement and works councils can be sent those links rather than waiting on a questionnaire.

An ATS is a system of record: it stores applications and tracks who is at which stage, and the work of screening, interviewing and assessing happens in other tools around it. HRBlade does that work. You can keep an ATS and use HRBlade for the hiring itself, or replace both.

Yes. There is an open REST API, webhooks and event streaming, and CSV/JSON export, and Enterprise includes custom integrations with systems such as Workday and SAP SuccessFactors. Running HRBlade in front of an existing ATS and syncing outcomes back is a common arrangement in mid-market and enterprise.

We migrate them. Open roles, candidate records, notes and templates come across as part of onboarding, which is the two-to-four-week window for a full replacement. You are not asked to reformat a spreadsheet.

Knowledge base and email support on every plan, priority support from Premium, a named Customer Success Manager on Premier, and a dedicated account team with solutions engineering, onboarding, team training, custom SLAs and quarterly business reviews on Enterprise.

Enterprise contracts carry a 99.95% uptime SLA along with custom terms negotiated with your procurement team. Self-serve plans run on the same infrastructure without a contractual SLA attached.