How does AI interview scoring work?

Iris assesses each answer against the competencies defined for the role, not the resume's keywords, producing an individual score per competency plus an overall interview score. Communication is scored as its own dimension, and every candidate is judged on one rubric so scores compare across the whole pool. The full scored read lands within minutes of the call ending, and every number links back to the moment that produced it.

Can AI run the whole interview, or does a person have to be there?

Both. In Autonomous mode Iris asks the questions, follows up on the answers and runs the whole interview with no human touch until the shortlist. In Human-led mode you or our recruiters run the room while Iris joins the call and takes the notes. You can switch per role, per round, or mid-hire, and every candidate lands the same scored dossier on the same rubric. We don't just score the candidate. We score the interview, whoever ran it: human-led rounds are scored on the same rubric as AI-led ones.

What is included in the interview scoring breakdown?

Every interview produces the same dossier: competency scores with a skills-gap read, a standalone communication score, a sentiment arc tracking confidence and stress from opening to close, AI-assist and integrity signals, written strengths and gaps, and a chaptered replay with a timestamped, searchable transcript. Every score links back to the minute of the interview it came from.

Can AI detect answers that are scripted or AI-generated?

Yes. Candidates now generate answers live with AI or read scripted responses. Iris assesses every response for authenticity: AI-generation likelihood per response with a confidence read, unnatural pacing and scripted cadence, response depth and consistency, and whether the interview answers match the resume's claims. These are signals surfaced for human review, never an automated rejection.

What video platform do the interviews run on?

Every interview, AI-led or human-led, runs on Team Connect, Teamified's own AI-native video platform. It works like the meeting tools you already know, except the agent is native to it: Iris can open a room, join the call and run the interview inside the software itself, with no bots dialling in or third-party recorders. Because the platform is ours, the recording, chapters, transcript and scores land straight in the same hiring pipeline the moment the call ends.

How do AI and human interviews save recruiter and hiring-manager time?

Two ways. First, confidence in the score: because every interview is scored on the same rubric with the evidence linked, a recruiter or hiring manager can trust the number without re-running the round. Second, chapters: instead of watching an interview end to end, they jump straight to the areas of concern the score points at, check the moments that matter, and move on. Reviewing a round takes minutes, not the length of the interview.

How soon after the interview do results come back?

The full scored read lands within minutes of the call ending. Instead of running a first-round call, a hiring manager can watch a chaptered replay jumped to the key moments and know where a candidate stands in about five minutes, before the meeting.

Does AI make the hiring decision?

No. AI does the work and humans decide. Every dimension Iris reports is a signal for a person to weigh, nothing is rejected on a score or a flag alone, and a hiring manager makes every call with the replay and the reasoning in front of them.

Iris · The interview agent

AI & human interviews. One rubric.

Iris runs the interview, or a person does, on Team Connect, our own AI-native video platform. Either way the candidate lands the same: recorded, chaptered and scored on one rubric, with the full read back within minutes of the call ending.

AI-led by Iris, human-led by your team or ours, or both on the same role
Every interview runs on Team Connect and lands scored on the same rubric
Visual, communication and skill signals recorded through the whole session
The full read back lands within minutes of the call ending

Explore the platform

A human always decidesConsistent by designPrivate by default

REC · AI-LED

teamified · interview room · candidate 04 · founder

Private by default

FOUNDER · CANDIDATE 04

“The alerting was mine. The dashboards were already there when I joined.”

12:43 / 14:26

VERIFIED REAL

Eye contact · good

Integrity check

Likely authentic

AI likelihood

Low

Natural flow

Yes

Confidence

80%

Subject scores

6 of 6

System design8.4

Communication8.2

Debugging9.0

Ownership8.8

One rubric. Every candidate.

Video chapters

00:42Role and scope

03:10System design

06:24Debugging a live incident

09:05Ownership

Jump straight to the evidence.

Delivery and behaviour

Attention: highly engagedPace: steady

Relaxed but attentive. Nods and hand gestures through the technical answers.

Evidence at 05:51

Sentiment arc

OPENCLOSE

Shortlisted, with the evidence attached

A human always decides. Sarah, 4:12pm.

Advance

See it run

One interview, end to end

From the question set to the scored dossier: watch how an interview runs on Teamified, and what lands in front of the hiring manager when it ends.

What are Teamified AI & human interviews?

Interviews on Teamified run on Team Connect, our AI-native video platform, whether Iris the interview agent runs the round or a person does. Iris asks job-specific questions autonomously or joins your human-led round, recording visual, communication and skill signals through the whole session. Every candidate is scored on one rubric, with the full read back within minutes of the call ending, and a human always decides.

Interviews

Iris

Asks, watches and reports

“I ask what you'd ask. I just never forget the answer.”

Your process, your dial

Can AI run the interview, or do I run it? Both, and the evidence looks the same either way.

Some teams want the whole round to run itself. Some want a person in every room. Most want both, depending on the role. So choose, per role, per round, or mid-hire.

Autonomous

Iris asks the questions, follows up on the answers and runs the whole interview. Zero human touch until the shortlist.

Candidates interview any hour, no scheduling. Your team wakes up to scored dossiers, not a queue of calls to book.

Human-led

You or our recruiters run the room. Iris joins the call, prompts the next question and records every signal.

The rooms that matter stay yours. Iris preps resume-specific questions and takes the notes, so every screen stays senior-quality.

Both

Autonomous for volume rounds, human-led for the roles that need a person. Switch per role, per round, or mid-hire.

Either way, every candidate lands the same: a recorded, scored dossier on the same rubric. The mode changes. The evidence does not.

Your process, your dial

Three routes in. One dossier out.

rubric v3 · locked

Autonomous

Iris asks, follows up and scores

Human-led

You run the room, Iris takes the notes

Both

Switch per role or per round

Candidate dossier

identical either way

Recording42:10

Criteria scored8 / 8

Claims verified6 / 6

Overall84

The mode changes. The evidence doesn’t.

AUTO · HUMAN · BOTH

One rubric, whoever runs the room

AI-led or human-led, every interview lands scored the same way.

The choice between an AI interviewer and a human one should be about the role and the round, not about which produces better evidence. So it never is. Iris-led rounds, human-led rounds and joint rounds all land in the same pipeline, scored on the same rubric, with the same dossier at the end.

That means results stay comparable across the whole pool: a candidate who interviewed with a person sits next to one who interviewed with Iris, on the same scale, with the same evidence attached. Choosing the human path never costs you signal.

What stays identical, whoever asks the questions

A Teamified team member at their desk, framed webcam-style during an interview — One candidate, one rubric

The rubric

Competency, communication and integrity scored on the same scale, AI-led or human-led.

The dossier

Chaptered replay, timestamped transcript, written strengths and gaps. Same format, every time.

The pipeline

Recordings, scores and transcripts land in the same place, feeding the same shortlist.

The decision

A person weighs the evidence and makes the call. Nothing is rejected by a machine.

The platform underneath

Every interview runs on Team Connect, our own AI-native video platform.

The agent is native to it: Iris opens the room, joins the call and runs the interview inside the software itself. Because the platform is ours, the recording, chapters, transcript and scores land straight in the hiring pipeline the moment the call ends.

Native, not bolted on
Iris runs calls in the software itself; human interviewers use the same rooms.
One pipeline, no exports
Recordings, transcripts and scores flow straight into the candidate's dossier.
Browser-based for candidates
Click a link and join. No downloads, no accounts, any hour that suits.

Explore Team Connect

Team Connect: a live interview room with two participants and the Meeting Copilot panel suggesting interview questions

The scoring breakdown

What lands in the dossier after the call?

Every interview, AI-run or human-run, produces the same six dimensions within minutes of the call ending. Pick one to see it.

Skills and competency scoring

Every competency scored from the interview evidence

Each answer is assessed against the competencies defined for the role, not the resume's keywords. Iris produces individual subject scores for the skills the job actually needs, plus a skills-gap read: where the candidate is strong, and where the evidence ran thin. One rubric, every candidate, so scores compare across the whole pool.

Individual score per competency, with the reasoning attached
Skills-gap analysis against the role's requirements
Overall interview score combining every dimension
Comparable across every candidate in the role

Every dimension is a signal for a person to weigh. Nothing here rejects a candidate on its own.

Interview summary · subject scores

Nikita S. · Back End Developer

Solid grasp of backend development concepts in .NET Core, Azure and databases. Communication pace could improve; presented professionally with specific examples.

Subject scores

Overall 7.0 / 10

Communication

6/10

Technical knowledge (.NET Core)

7/10

Technical knowledge (Azure)

7/10

Database knowledge

7/10

Professionalism

8/10

Where the time comes back

Stop watching interviews. Start checking moments.

The old way of reviewing a round is sitting through it again: an hour of video per candidate, or a first-round call you run yourself. Here, the score does the sitting through. You trust the number because every one links to the moment that produced it, and when something needs your eyes, the chapters take you straight there.

Confidence in the score

One rubric across every candidate, every score linked to its evidence, every analysis bias-checked by FairLens. You do not need to re-run the round to believe the result.

Chapters to the concern

The dossier points at the weak spots; the chapters jump you to them. Watch the system design answer and the hesitation, not the pleasantries.

Minutes, not hours

A hiring manager can know where a candidate stands in about five minutes: read the dossier, check the flagged moments, decide. Part of why the whole process runs a minimum of 50% faster.

Trust the candidate again

Can AI detect scripted or AI-assisted answers? Ours flags what does not add up.

Candidates now generate answers live with AI, read scripted responses through earpieces, or rehearse coached segments word for word. A note-taker summarising the interview you already sat through cannot catch it. Iris watches the whole recording, speech and video, and flags what does not add up.

AI-generation likelihood assessed per response

Scripted cadence and unnatural phrasing identified

Interview answers cross-checked against the resume's claims

Confidence drop-offs flagged where coached segments run out

What Iris flags, and what happens next

AI-assist signals

Per-response likelihood with a confidence read. Elevated scores are flagged, never actioned.

Speech pattern anomalies

Unnatural pacing, scripted cadence and memorised-sounding responses called out with timestamps.

Resume contradictions

Claims in the CV cross-checked against what the candidate actually said on camera.

A person makes the call

Every signal lands next to the replay for your hiring manager to judge. Iris never rejects anyone.

Iris

FairLens

Every Iris analysis is checked by FairLens

A second agent screens every interview analysis for bias, and any hiring manager can re-run the same interview through a different model without leaving Teamified.

Trust, bias & fairness

People decide

The evidence lands with a person, not a verdict

Minutes after the call ends, the recording, transcript and scored dossier are in front of a real reviewer on your team. The AI surfaces the signals. A human always decides.

Every competency score linked to the moment it came from
Chaptered replay and transcript ready when the call ends
No candidate is rejected without a person saying so

Interview ended

Recording + transcript ready

Competency scores

6 dimensions scored

Each linked to its moment

A human always decides

The rest of the platform

Job creationMason + Quill + Compass

Job posting & sourcingPostman + Hunter

Screening & verificationSift + Vera

Trust, bias & fairnessFairLens

Scheduling & candidate commsCal + Echo Where the six weeks of a traditional hire goesThe story For job seekers: feedback on every interviewThe candidate side Your process & pipelineYour rules

One platform, three missions, one reason

Delivered by our AI agents, our people and the platform.

01 · Agentic AI hiring platform

Hire high quality, a minimum of 50% faster

Read our missions

Questions, answered

AI & human interviews, answered.

Direct answers to the questions teams ask about AI and human video interviewing and scoring.

The agent platform

Alexia, the platform every agent runs on

Every agent you meet on this site runs on Alexia. Each one is configured with a primary model and a secondary failover, and each connects to more than 1,000 software products. They range from agents built for a single task to agents that help run whole teams: gathering data, setting goals, creating tasks, and managing work and group chats.

Want to build your own agents? Log into Alexia and start today.

Build your own agents

Runs