Asked often. Answered plainly.

Pyro is an applicant tracking system built for staffing agencies and startups, with AI in the process rather than bolted on the side. It writes down what good looks like for a role once, then holds the posting, the search, the screen, the first-round interview and the human round to that standard. Every decision along the way is written to a record that shows its work.

Pyro is the system of record. Roles, postings, candidates, applications, interviews and the audit record all live in it. You do not have to run it alongside another ATS, but you can if you wish to: it integrates seamlessly.

When you write a role, Pyro turns the essentials into a short list of criteria, typically five to ten. The search ranks against that list, the screen scores against it with a quoted line for each criterion it can evidence, ARIA asks about the ones it cannot, and the human round closes what is left. Nobody has to guess what “good” meant, because it was written down before anyone was assessed.

No. Connect the shared drive your CVs already live in and Pyro reads them where they are. The top match in your first search is usually someone who applied years ago and was missed on a keyword.

English today. Postings, search, screening and ARIA interviews are all in English. We will say when that changes rather than list languages we have not tested.

Yes. Client-posting mode lets an agency write, publish and run a role on a client's behalf while keeping the client's candidates and record separate from your own.

No. ARIA runs a spoken first round and updates the evidence for each criterion. Advancing, rejecting and hiring are done by a person, and the record names that person. Pyro is built so that a human decision is recorded on every hiring outcome; that is a design constraint, not a setting.

Yes, before the interview starts. ARIA is introduced as what it is, and the candidate signs a consent that names the scope, which is typically screening plus the ARIA round. That consent is a row in the record, with who signed it and when.

They can. Declining does not change the evidence they already have; it leaves the open criteria for a human round instead. What happens next is the employer's call, and we recommend offering a human first round to anyone who declines.

ARIA asks for specifics a rehearsed answer rarely has: what the number was before, what it was after, and what the candidate personally changed. A general answer gets a narrower question, and the score moves only when the answer is specific enough to quote with a timestamp. It is not a lie detector. It is an interviewer that keeps asking.

ARIA rounds are transcribed. The transcript, the timestamps and the evidence drawn from it are part of the application record, which the employer keeps and the candidate can see. Candidates are told this before they start.

It sits beside the interviewer and is visible only to them. It keeps the question list, retires questions as they are asked, drafts a follow-up from the candidate's last answer, and tracks which criteria are now evidenced. It never speaks to the candidate and it never decides.

Foundation models from Google Cloud and Anthropic, chosen per task. We do not publish how tasks are routed, but every AI run writes the model and version it used to the record, so you can always see which model touched a decision.

No. Your CVs, transcripts and records are used to serve your account and nothing else. We do not use them to train models for other customers, and our AI providers are contractually barred from training on what we send them.

On Google Cloud. Production data is stored in Google Cloud's India region today. EU data residency is available on Enterprise plans, with a Data Processing Agreement that names the region and the sub-processors.

At the database. Every tenant's rows are isolated by row-level security enforced by the database itself, not by application code that could forget to filter. Personal data is encrypted at rest with per-tenant keys, and erasure requests execute as deletions at the database rather than as a hidden flag.

Not yet. We would rather say that plainly than imply it with a badge. The controls that matter are enforced at the database and written to the record, and we will walk your security team through them on a call.

Four ways, all enforced rather than advised. A posting that contains a biased line cannot be published, and no role can override the gate. Protected characteristics such as gender, ethnicity, age, disability and veteran status are never shown to anyone making a decision and never used by the AI. Scores move only on quoted evidence, never on inference. And every hiring decision records the person who made it.

Yes, and the details are on our privacy page. In short: candidate consent is scoped and revocable, erasure is a real deletion, human oversight is recorded on every hiring decision, and every AI run is logged with its model and version. A Data Processing Agreement is available for customers.

The same evidence the recruiter sees: which criteria your application evidences, which are still open, what stage you are at, and what changed after your interview. When somebody looks at your application and moves it, you are told.

No. An unanswered criterion on Pyro is a question, not a rejection, and the decisions to advance, reject or hire are made by a person at the employer. The record names who.

Yes. You can withdraw the consent you signed from your portal at any time, and you can ask for your data to be deleted. Where the employer holds your application, we pass the request to them and help them act on it. Where we hold your global profile, we delete it ourselves.

No. One profile follows you to every role on Pyro. Edit it once and every application you have open sees the change.

You pay for usage and for success, not for seats. Users are unlimited on every plan. AI runs and ARIA rounds are metered, and your usage is always visible in the application, so the bill is never a surprise. Success is billed on hires. Talk to us for numbers on your volumes.

Book a demo and bring one open role. We run the whole line on it, on the drive you already have, in front of you. If it holds up, you connect your drive and write your first standard the same day.

A role can be written, searched and screened on the first day, because there is nothing to migrate. Connect the shared drive and the CVs are already there.