Before you begin
Praxel Hire is a workspace tool for recruiters and hiring managers. Praxy drafts briefs, ranks candidates, and runs first-round voice interviews. You decide every consequential action.
A work email, a quiet room for the candidate side when you test it, and 10 minutes to walk through your first role.
You already have a rough job description in your head. Praxy will extract the structured version through conversation.
Praxel runs in the browser. Candidates join interviews from a link — no downloads, no accounts.
Create your workspace
Every team lives inside a workspace. You'll create one on first sign-in. Teammates join the same workspace through an invite.
- 01
Sign up
Head to Create your workspace. Sign up with work email or Google. If your company already uses Praxel, ask a teammate to invite you instead — don't create a parallel workspace.
- 02
Name your workspace
After sign-up, Praxy asks for a workspace name. This is your company or team name. It shows up in candidate-facing emails, so use something candidates will recognise.
- 03
Land on the dashboard
First view shows zero jobs, zero candidates. Praxy greets you by first name. From here the next move is almost always Draft a role.
Describe the role in plain English. Praxy turns it into a structured brief.
Draft a role with Praxy
Intake is a short conversation. You describe the role; Praxy extracts a structured brief in real time on the right side of the screen. You refine by talking back. The brief becomes the job's source of truth for screening and interviews.
- 01
Open intake
From the dashboard or the Jobs page, click New role. You land on
/intake. - 02
Describe the role in your own words
Type how you'd describe the role to a friend. A few sentences is enough. Example:
We need a senior backend engineer for our payments team. Python, 6+ years, comfortable with Postgres and event-driven systems. Remote in India. Budget around 45–70L.
- 03
Watch the extraction fill in
As you type, the right panel fills: title, skills, seniority, location, years, salary range, description. If something's wrong, say so in plain English — "make it remote-first" or "drop the Postgres requirement." Praxy re-extracts.
- 04
Create the job
When the extraction looks right, the Create job button turns active. Click it. You land on the new job's overview page. The job starts as draft — only visible inside your workspace until you invite candidates.
- Title
- Senior Backend Engineer
- Skills
- Python · Postgres · Event-driven
- Seniority
- Senior
- Location
- Remote · India
- Experience
- 6+ years
- Salary
- ₹45L – ₹70L
Source candidates
Two ways candidates land in a job: you add them manually, or you run a LinkedIn sourcing pass. Both end up in the same candidates table, ranked by screening score.
For people already in your pipeline (referrals, inbound). Job page → Candidates tab → Add candidate. Name is required; email and LinkedIn help, resume text is optional.
Job page → Sourcing tab → Start run. Praxy uses the brief to query LinkedIn, pulls matching profiles, and drops them into this job. Runs take 1–5 minutes.
- 01
Start a sourcing run
On the job's Sourcing tab, confirm the prefilled keywords, seniority, location, and how many results you want (default 25). Click Start run. Status shows running.
- 02
Wait for results
You don't have to sit on the page. When the run finishes, new candidates appear on the Candidates tab with status pending. Praxy begins scoring them against the brief automatically.
- 03
Review the ranked list
Candidates sort by screening score. Open one to see the full profile, resume pull, and Praxy's screening notes. You can disagree — scores are a draft, not a verdict.
Design the interview
Every job gets an interview plan and an interviewer voice. You set both once per role. All candidates for that role get the same plan and the same voice — so you're comparing like with like.
4a · Approve the interview plan
Open the job → Plan tab. If no plan exists, click Generate plan. Praxy drafts five sections from the brief:
- PersonaHow the AI interviewer behaves — tone, pace, depth.
- Evaluation criteriaWhat signals we're listening for. Drives the score.
- StructureFlow of the 10-minute conversation.
- System promptInternal instructions to the AI. Advanced users only.
- QuestionsOrdered question list with follow-up guidance and required flags.
Edit any field. Reorder or toggle required on questions. When happy, click Approve plan. Only approved plans can be used in interviews — this is the gate.
4b · Pick the interviewer voice
On the job's Overview tab, the Interviewer voice card lists 11 curated voices. Each has a preview button — click to hear a short sample. Pick one that fits the role's tone (warm for customer-facing, crisp for engineering, etc.). Selection saves immediately.
All interviews for this role use this voice.
Invite candidates
Once the plan is approved and you've picked a voice, you can invite. Each candidate gets a unique, time-boxed link. Links expire after 7 days by default.
- 01
Invite from the candidates table
Click Invite on any row. Praxy creates a fresh interview record and signs a JWT token — this becomes the candidate's link.
- 02
Send it
If the candidate has an email on file, Praxy emails them automatically. You'll see Email sent confirmed in the interview link card. Either way, the raw link is visible — copy and share over WhatsApp, Slack, or manual email if you prefer.
- 03
Re-invite safely
Clicking Invite again before expiry re-uses the same link (won't create a new interview). After 7 days or after a completed interview, a new Invite creates a fresh token.
Sent to aarav@example.com · Expires Apr 26, 2026
https://hire.praxy.me/room/eyJhbGciOiJIUzI1NiIsInR5cCI6Ikp…What candidates see
Worth knowing so you can answer questions on the phone. The candidate's side is intentionally short and distraction-free.
- 01
Land on the room page
Candidate clicks the link. They see a greeting with their first name, your workspace name, the role, and a microphone test. No login. No download.
- 02
Pass the tech check
Browser asks for mic permission. The page shows a live audio level so they know it's working. If the mic fails, they see clear guidance (usually a browser permission toggle).
- 03
Join the live interview
They click Join. The AI interviewer greets them and walks through the approved plan. Typical length: 8–12 minutes. Transcript and audio are recorded.
- 04
Land on the thank-you page
When the interview ends, the page confirms completion and tells them the team will follow up. No score is shown to the candidate.
You've been invited to a short voice interview for the Senior Backend Engineer role. Find a quiet spot, put on headphones, and click join when you're ready.
Review interviews
Completed interviews show up on the candidate's profile page and in the job's Interviews tab. You get the recording, the full transcript, an AI summary, and an integrity score.
- 01
Open the candidate
From Jobs → candidate row, or from the dashboard's "Latest scored" card. The profile page stacks the latest interview on top, older ones below.
- 02
Watch or skim
Use the recording player for a quick listen. Use the transcript to search for specific keywords (say, "Postgres" or "team size"). The summary is the fastest read — but don't trust it blindly; it's a draft.
- 03
Check integrity
Every interview has a fraud score. Low scores are usually clean. Mid and high scores come with a list of specific signals — see the next section.
- 04
Make a call
Nothing auto-advances or auto-rejects. You decide whether to move the candidate forward based on what you see. Praxel does not hide reasoning — every score has the evidence attached.
Integrity & fraud signals
Voice interviews invite a particular kind of cheating — tab-switching, reading from ChatGPT, someone else taking the call. Praxy runs an independent integrity pass on every completed interview and flags specific evidence. Never on protected attributes (accent, gender, appearance) — only on behavioural signals.
The three buckets
Likely clean. Default.
Mixed signals. Spot-check before advancing.
Strong evidence of cheating or impersonation. Review carefully.
The Integrity hub
Head to Integrity in the left nav to see every flagged interview across your workspace — not just one role. It's tabbed by the three buckets. Start with Flagged before shortlisting.
What a signal looks like
Signals are specific, time-stamped, and carry a confidence score. Examples you'll actually see:
- Long silent gaps before technical answersCandidate paused 18s before answering a basic question, consistent with reading from a second screen. Confidence 0.74.
- Voice characteristic shift mid-interviewPossible second speaker after 06:22. Worth verifying on the recording. Confidence 0.61.
- Phrasing resembles LLM outputThree consecutive answers match the structure and cadence of a generic GPT response. Confidence 0.58.
FAQ
Yes. Edit the plan at any time; the next invited candidate gets the new version. Already-completed interviews keep the plan they were run against.
Links expire 7 days after they're sent. A candidate can join multiple times within that window if they drop.
No. Interviews are voice-only by default. A laptop or phone with a mic is all they need. Video avatar is an optional per-workspace feature.
Spoken English is the default. Reach out to hello@praxy.me if you need another language — we maintain a curated list per region.
Yes. Everyone in the workspace sees every candidate. There's no per-role access control today.
Candidates can be deleted from their profile page. Interviews are retained for audit purposes; contact support if you need a specific recording purged.
No. Candidates and transcripts stay in your workspace. We never train models on customer data. See the Responsible AI page for the full stance.
Get help
Email us and we'll get back within one business day. If it's urgent, mention it in the subject line.