ProBB AI checks each resume against the job, runs a short AI voice interview, and flags what you should look at before the candidate goes to your client.
“Strong backend experience” can’t be checked against a resume. Name the stack and the level.
Strong backend experience
4+ years building and running Go or Java services in production
DismissApply
What screening will look for
MustGo or Java servicesPayment APIsOn-call ownership
NiceCard networks
7 checks passed
Fig. 1 · JD coach · app.probbai.com/job-descriptionsSample · fictional data
02
Drop in the resumes
Upload them in bulk. Each resume is checked against the job's requirements, with the supporting evidence and the gaps listed for every criterion.
Reports
Review your team’s candidate analyses and interviews.
Search candidate, role, email, or job description
All fit levelsNo grouping
Priya Nair
For Senior Backend Engineer
8.4/10Overall evidence
Final
Evidence summary
Owned Go payment services end to end, with a measured, zero-mismatch ledger migration.
Interview
Proceed88% readiness
Proceed to a final conversation about system design trade-offs.
Confirm how on-call scaled as the team grew.
2 hours ago
By Sana Iyer
Review evidence →
Aarav Mehta
For Senior Backend Engineer
2 flags
7.4/10Resume fit
Provisional · interview pending
Evidence summary
Specific Go and payments evidence; the Paycore and Finlytics dates overlap.
Interview
Interview pending
2 hours ago
By Sana Iyer
Review evidence →
Fig. 2 · Reports · app.probbai.com/reportsSample · fictional data
03
Invite to an AI interview
Invite one candidate or a whole shortlist. They take a voice interview of about 15 minutes in their browser, with questions based on their resume and your job.
Fig. 3 · Live AI interview · app.probbai.com/interviewSample · fictional data
04
Share the report
You get a report with the transcript, strengths, gaps and follow-up questions, and you choose which parts to share with a client or hiring manager.
Hiring manager debrief
Priya Nair
Senior Backend Engineer
Share
8.4/ 10
Overall evidence
Strong Evidence
Final
Executive Summary
Owned Go payment services end to end with measured results. Interview answers were specific and consistent with the resume.
Fig. 4 · Share dialog · app.probbai.com/reports/priya-nairSample · fictional data
01
Add the job
Paste a job description. The JD coach scores it and suggests fixes for vague or missing requirements before you screen anyone.
02
Drop in the resumes
Upload them in bulk. Each resume is checked against the job's requirements, with the supporting evidence and the gaps listed for every criterion.
03
Invite to an AI interview
Invite one candidate or a whole shortlist. They take a voice interview of about 15 minutes in their browser, with questions based on their resume and your job.
“Strong backend experience” can’t be checked against a resume. Name the stack and the level.
Strong backend experience
4+ years building and running Go or Java services in production
DismissApply
What screening will look for
MustGo or Java servicesPayment APIsOn-call ownership
NiceCard networks
7 checks passed
Fig. 1 · JD coach · app.probbai.com/job-descriptionsSample · fictional data
03Integrity checks
Signals worth a second look.
ProBB AI also notes things in the resume or the interview that you may want to check yourself.
Resume signals
Keyword stuffing, hidden text, timeline inconsistencies and claims without supporting detail are noted for review.
Skills
Python · Airflow · dbt · Snowflake · Spark
expert senior staff principal python spark kubernetes ml ai llm
Hidden text possibleKeyword stuffing
Also checked: timeline inconsistency · unsupported claim density · duplicate content · formatting anomaly.
Speaker continuity
If more than one voice seems to be answering during the interview, the report mentions it so you can follow up.
Voice 1
Voice 2
Speaker checkConsistent
Claims that need a follow-up
Answers that may not hold up technically are marked as questionable, with a short, neutral reason.
Interviewer
How did you avoid duplicate payments when consuming from Kafka?
Candidate
Kafka gives exactly-once delivery end to end by default, so duplicates were never a concern for us.
QuestionableExactly-once needs idempotent producers and transactional reads; worth a follow-up question.
Answers that miss the question
Generic or rehearsed answers that don't address the question are noted.
Interviewer
Tell me about a time you pushed back on a technical decision.
Candidate
I'm a strong team player and always focus on delivering value to customers through collaboration and best practices…
Didn't address the question
Integrity checks are advisory and never reject a candidate. A person makes the call.
For your candidates
Fair to the people being interviewed.
Your candidates are your reputation. Here's what the AI interview looks like from their side.
Consent comes firstNothing is recorded until the candidate reads the notice and agrees. If they decline, no assessment is created.
About 15 minutes, in the browserNo app to install. The interview never runs past 22 minutes, and they can stop at any time.
Scored on what they say, not how they say itAccent, grammar, pauses, nervousness and non-native phrasing aren't penalised. Protected characteristics, emotional state, voice stress and voiceprints aren't used to score anyone.
A person stays reachableCandidates can ask the recruiter for human review, a reasonable accommodation or an alternative assessment.
This structured interview targets 15 minutes, may use additional recovery time, and never exceeds 22 minutes.
The conversation will be transcribed and audio recorded for hiring review. The recording is available only to authorized members of the hiring organization.
I have read the notice and consent to recording, transcription and AI-assisted evaluation for this interview.
Read T&C and privacy notice
Start interview
Fig. 5 · Before the interview · app.probbai.com/interviewSample · fictional data
04The report
The report your client or hiring manager reads.
The same layout as the report in the app, shown here with a fictional candidate. Switch tabs to see each part.
01An overall score with the reasoning written out
02Marked final or provisional, so a resume-only score isn't mistaken for a final one
03A short executive summary at the top
Share any report with a link you control
Hiring manager debrief
Priya Nair
Senior Backend Engineer
Interview: Completed
Review interview
Candidate context
Extracted identity and resume signals used for this debrief.
priya@example.com 6 years Bengaluru, India Backend Engineering
8.4/ 10
Overall evidence
Strong Evidence
Final
Executive Summary
Priya has owned Go payment services end to end, including a ledger migration with a measured, zero-mismatch cutover. Her interview answers were specific and matched the resume. Explore how she would scale on-call practices as the team grows.
Overall Assessment
Combined resume, JD, and interview evidence
Final
Proceed to a final conversation about system design trade-offs.
8.4/10
Overall evidence
8.1/10
Resume–JD baseline
85%
Interview coverage
82%
Assessment confidence
Fig. 6 · Candidate report · app.probbai.com/reports/priya-nairSample · fictional data
How the resume score is built
Each resume is scored against the job on five criteria, each with written reasoning, the evidence found and the gaps that remain.
01Role requirements match
02Technical evidence
03Experience scope
04Execution impact
05Evidence risks
How the interview score is built
Every interview finding cites the answer it came from. Communication is scored leniently: accent and non-native phrasing aren't penalised.
Core role skills35%
Applied problem solving25%
Role and experience fit25%
Communication clarity10%
Evidence coverage5%
Matching
Your talent pool, matched both ways.
Find the best people for a new req from candidates you've already screened, or the best open roles for someone on your bench.
Requirements are split into must-have, important and optional
Transferable experience is called out, not hidden
Shortlist, dismiss or run a full analysis in one click
Matches
Find candidates for a job or jobs for a candidate.
Candidates for a jobJobs for a candidateSaved matches
Job and filters
Senior Data Engineer
214 eligible candidates · Data platform
Find matching candidates
All match levelsAll decisions
#1
Lena Fischer
Report readyStrong potential
Data Engineer
7 of 8 criteria supportedBerlin
Reviewing evidence
#2
Rahul Verma
Possible match
Analytics Engineer
5 of 8 criteria supportedPune
Review evidence
#3
Sofia Marin
Exploratory match
Backend Engineer
4 of 8 criteria supportedLisbon
Review evidence
Fig. 7 · Matches · app.probbai.com/matches/candidatesSample · fictional data
Who it's for
Two ways teams use it.
You
Recruiters & agencies
Screen your candidate bench against each new req, and send clients a report instead of a bare resume.
Upload resumes, start reports and invite candidates over a REST API, then get signed webhooks when results are ready. Available on request, with a sandbox to test against.