A narrated video walkthrough of every feature, from the first interview to the final report
Watch one hire from start to finish: creating an interview with AI, Job Intelligence, inviting a candidate, the candidate's interview, the report, the candidate intelligence page, the Big Five personality pilot, practices, and team settings. Jump to any chapter below, or turn captions on in the player.
Hiring often starts from a CV, and a CV is a list of claims. IntervAI turns a job description into structured requirements, runs AI interviews, and shows what was actually measured.
This walkthrough is for recruiters and hiring teams who run interviews, and for candidates who take them. Everything you'll see is the real product, running locally with fictional data.
We'll follow one hire from start to finish: a Senior Backend Engineer role at a fictional company, Northwind Freight, and a fictional candidate, Alex Morgan.
To get started, open your IntervAI address and create an account with your email and a password of at least eight characters. You'll confirm your email with an eight-digit code.
After that, sign in with your email and password. If you forget it, the reset flow sends you a code.
The first visit opens a short guided tour. You can take it now, or pick it up later from the compass icon.
The sidebar holds everything you'll use: interviews, practices, sessions, the question library and your answer bank. Project settings, support and usage sit at the bottom.
At the top, the breadcrumb switches between organizations and projects. There's quick search, a dark mode toggle, and the dashboard shows your activity at a glance.
Let's create our first interview. Click New Interview. The AI generator turns a plain-language request into a complete interview.
Describe the role and what you want to assess. You can start from a template, attach a job description as a link or PDF, and add a resume.
Set the duration, the number of questions, the language, the interviewer's tone and how deeply it follows up. Then choose the channels: chat, voice, or video.
Simulation mode goes further: instead of only asking questions, the AI builds workplace scenarios where the candidate does the job. We'll keep it off for this role.
Press Generate. The AI writes the questions and the assessment criteria.
Review the draft. Edit or remove any question, add your own, or import questions from other interviews. If the direction is off, describe the change and refine it with AI.
When it looks right, choose Accept and Create. Prefer full control? The Manual tab lets you build an interview from scratch.
The Content tab is the question builder. Reorder questions, edit them, and set the assessment criteria the report will score against.
Each question has a type: open-ended, single or multiple choice, coding with a code editor, a whiteboard for diagrams, research questions that probe deeper, and roleplay scenarios.
On coding questions, candidates write in a full code editor and the AI reviews the code. In this version the code is saved and reviewed, not executed.
The Settings tab controls access and behaviour. By default an interview is invite-only. Create a shareable link if anyone with the link should be able to start.
Here you also edit the title, description, objective and time limit, and choose the channels. Video needs voice.
Anti-cheating mode makes camera, microphone and screen sharing mandatory, tracks tab switches, blocks pasted text from outside, and warns about extra monitors.
Finally, give the AI interviewer a name, a tone, a follow-up depth and a language, and save. The Share button copies the link, or lets you preview the interview as a candidate before anyone else sees it.
Now the intelligence layer, SYRA. Because we attached the job description, the Job Intelligence tab has already analysed it into structured skills, competencies and requirements. Edit the description here and re-analyse whenever it changes.
Every skill keeps the line of the job description it came from, and is marked required or preferred. If AI isn't available, a rule-based fallback runs and says so.
These levels are machine estimates. Review them: adjust a level, mark a skill required, add or remove skills, competencies, requirements and responsibilities.
Save the review. Until a recruiter does, every report says the requirements are not reviewed. And if the job description changes later, SYRA flags the analysis as stale.
Next, invite candidates. In the Sessions tab, Add offers three ways: one by one, an Excel import using the template, or a batch of PDF resumes the AI reads for you.
We import Alex Morgan's resume. The AI extracts the details, and the resume itself is kept for the candidate's profile. Every candidate gets a personal invite link. Copy it and send it however you like.
The table tracks each session: not started, in progress or completed. Search, filter, open reports, reopen a session, or export the list to Excel.
Here's what Alex sees. The invite link opens a welcome screen with a short guide. Alex confirms a name and email and begins.
Depending on the setup, a checklist asks for camera, microphone and screen permissions. With anti-cheating on, these can't be skipped.
In a chat interview the AI interviewer asks one question at a time and follows up on the answers. Everything you see here is a live AI conversation.
In voice mode the conversation is spoken, and candidates can switch between voice and chat at any time. Video adds camera and screen recording.
Coding questions open the code editor, and whiteboard questions open a drawing canvas the AI can see. Candidates can move to the next question, skip one, or go back.
Progress is saved as they go, so a dropped connection isn't lost. When Alex ends the interview, the session is submitted and a confirmation appears.
Back in the recruiter view, the completed session appears under Results with its status, duration and messages.
Open it. The report starts generating as soon as the session ends: the AI reads the whole transcript. For an interview still in progress, you can end it and generate the report yourself.
You get a summary, an average score, and a question-by-question evaluation with a score out of ten, strengths and areas for improvement.
Assessment scores follow the criteria you defined. Depending on the interview, the report also holds research findings, an integrity log, code snippets, recordings and the full transcript.
Export the report as a PDF to share it. Lists of interviews and sessions export to Excel.
The Sessions page lists every session across your interviews. From a completed session, create a Human Profile, the candidate's evidence-based profile.
The Human Profile separates claims from evidence. Skills from the CV are claimed: confidence is capped at point seven, and they never get a level. Interview answers are assessed, and the level comes from the scores.
The Competency Comparison puts every requirement next to what was measured: meets, exceeds, below, or insufficient evidence. Unknown is a real answer here, not a guess.
The assessment plan ranks what to check next, and explains why. The strategy groups those into packages you can launch.
Launching asks for confirmation and runs once per package. Nothing is scheduled automatically. Progress shows on the card, and in this version, results from launched assessments are not added to the profile, and the page says so.
The Human Digital Twin brings it together. Role fit is a category, never a score. Potential and career path show as unavailable, with the reason. It never calls AI, it lists its limitations, and it is not a hiring decision. The recruiter decides.
The Big Five personality questionnaire is a complete, working pilot. It is experimental and not approved, and it sits behind a server flag that is off by default. We switched it on for this local demo only.
It uses the IPIP-NEO-120: one hundred and twenty public-domain statements covering five traits: Openness, Conscientiousness, Extraversion, Agreeableness and Neuroticism.
The recruiter creates a private link and sends it to the candidate. It can be cancelled at any time.
The candidate sees twelve pages of ten statements, each answered on a five-point scale from very inaccurate to very accurate: how well the statement describes them. The items are in English only.
A page must be complete before moving on. Answers are saved, so the candidate can leave and come back with the same link. At the end, they submit.
Back on the candidate page, the recruiter sees the attempt as submitted and opens the results.
Each trait is a raw sum of twenty-four answers, from twenty-four to one hundred and twenty. Reverse-worded statements count in reverse. That's all it computes.
There are no norms, percentiles or high and low labels, no interpretation, and no hiring recommendation. Results are stored separately and are not used by the profile, the comparison or the Digital Twin.
Practices turn any interview into a coaching space, without creating real candidate sessions. On the interview's Practices tab, add context: the company, the role, the job description and resume notes.
Practice interview opens a focused session. The AI coach asks each question. You answer by voice, with a live transcript, or by typing.
Each answer gets a score out of ten, a verdict, strengths, improvements and missing signals. Try the question again, or open a suggested answer as a study aid and ask the coach to rewrite it.
The Practices dashboard tracks every run: averages, attempts and durations, with filters, export and cleanup.
Your strongest answers go to the answer bank, with notes, so you can review them before the real interview.
The Interviews page lists every interview you've built. Search, filter by date or access, duplicate one as a starting point, delete in bulk, or export the list.
The Questions page is your library: every question across interviews, searchable and filterable, ready to import into new ones.
Organizations hold your projects and your team. Add members by email and give them a role: owner, admin or member. Projects have their own settings and members.
For integrations, create API keys and use the public REST API. Its OpenAPI description is served by the app.
Usage shows interview templates, session time and seats over time. And in Account settings you can change your name, password and language, or delete your account.
Need help? The Support panel links to the documentation and lets you send a message to the team.
That's the full journey: a job description becomes reviewed requirements, an AI interview becomes a scored report, and the evidence becomes a profile, a comparison, a plan and a Digital Twin.
What's not here yet: the Big Five stays a pilot; launched assessment results don't reach the profile; interview code isn't executed; O-star-NET, the Cloud Human Avatar, webhooks and billing have no screens yet.
IntervAI and SYRA give you better questions, clearer evidence, and the final decision stays with you.