SuprHiro uses AI to take repetitive work off hiring teams. Decisions about people belong to people, so we have designed the product to keep them there — and where something does run automatically, we say so here. This page is for the employers who use SuprHiro and for the candidates they hire.
1. Our principles
- People decide. AI output is input to a recruiter’s judgement, not a substitute for it.
- Safe by default. Automations are off until a customer switches them on, and start in preview mode.
- Explainable scores. The match score is a simple, visible formula, not a black box.
- Transparency. Candidates are told when a call or an interview is recorded.
- No training on your data. We do not use customer or candidate data to train AI models.
2. Where AI is used
| Feature | What it does | Who acts on it |
|---|---|---|
| Resume parsing | Reads the resume and extracts contact details, skills, experience and education | Recruiters review the profile |
| Match score | Compares the resume's skills with the job — 70% weight on required skills, 30% on nice-to-have skills. AI writes only a one-line explanation | Recruiters, or automations the customer configures |
| AI phone screening | An AI assistant calls candidates to ask screening questions, and summarises the answers | Recruiters; see section 4 |
| AI interview notetaker | Transcribes and summarises an interview when switched on for that interview | The interview panel |
| Interview summaries | Summarises the panel's notes and feedback, including a suggested recommendation | The hiring team decides |
| Staff assistant | Answers questions and carries out tasks for staff; anything destructive asks for confirmation first | The staff member using it |
| Talent suggestions | Suggests past applicants to the same company for new roles | Recruiters decide whether to contact them |
| Attrition insights | Flags new hires who may be at risk of leaving, from onboarding progress and engagement signals | Admins and analysts review; never acted on automatically |
AI processing is carried out by the providers listed on our Sub-processors page, only to produce the result requested.
3. Automations the customer controls
Customers can set rules — for example, shortlist candidates who score above a threshold, or reject those below it. These rules:
- are off until the customer switches them on;
- start in preview mode, which records what would have happened without doing it;
- ask a person to approve score-based rejections, and notice-period and pay-based rules, by default;
- can be undone for 24 hours — the candidate’s stage is restored, although emails already sent cannot be recalled.
A customer can choose to switch preview or approval off. If it does, those actions run without a person approving each one, and the customer is responsible for them.
4. Steps that run automatically
These built-in workflow steps run without a separate approval:
- Two missed interviews. If an interviewer records that a candidate did not attend a second interview, the candidate is marked as rejected.
- Positive interview feedback. When every interview is complete and the panel’s feedback is “hire” or “strong hire”, the candidate moves to the offer stage. The offer itself is always prepared and sent by a person.
- Positive screening call. If an AI screening call ends positively, or the candidate asks for a callback, the candidate is shortlisted and sent a link to book an interview. An AI call never rejects a candidate.
- Filled roles. A job closes to new applications once all its openings are filled.
5. For candidates
- AI calls — the assistant tells you at the start that the call is recorded. If you would rather not continue, say so or end the call; you can ask the company to speak to a person instead.
- Recorded interviews — if a company uses the notetaker, you are told in your invitation and the notetaker announces itself in the meeting. Tell the interviewer, or reply to the invitation, if you do not want it recorded.
- Human review — you can ask the company you applied to for a person to review any decision about your application, and to explain the main reasons for it. If you cannot reach them, email hiring@supahiro.co and we will pass your request on.
- More about your rights is on our Your Data Rights page.
6. Known limits
AI can misread a resume, mishear a call, or summarise an interview inaccurately, and a skills-based score cannot see potential, transferable experience or context. Resumes that describe the same skill in different words may score differently. That is why the product is built for a person to review AI output, and why we ask customers to do so before acting on it.