What is interview intelligence? Definition, how it works and an example

Interview intelligence is software that captures what is said in a job interview and turns it into structured evidence for the hiring decision: a transcript, notes organized under the criteria the job is assessed on, and a draft scorecard that the interviewer reviews and signs off. It supports the person running the interview rather than replacing them, which is the main thing that separates it from an AI interviewer.

This page defines the term, walks through how it works with an example, and compares it with the terms it gets confused with. If you are choosing a product, the interview intelligence software buyer's guide has the vendor landscape and the questions to ask in a demo, so they are not repeated here.

What does interview intelligence actually do?

Strip away the marketing and every interview intelligence product does some version of the same five jobs. Products differ mainly in how they do the first one and how far they go with the last two.

1. Capture the conversation. Record or stream the audio of a video interview, a phone screen or, in some products, an in-person meeting.
2. Transcribe and label speakers. Turn speech into text and work out which lines belong to the interviewer and which to the candidate.
3. Structure the content against the interview plan. Sort what the candidate said under the competencies, scorecard questions or requirements agreed for the role, instead of leaving it in time order.
4. Draft the evidence. Produce notes, highlights and a draft scorecard in which each point can be traced back to the transcript.
5. Hand it on. Send the result where the hiring decision is made: the applicant tracking system (ATS), a debrief, a hiring committee packet or, for agencies, a client submittal.

Larger platforms add a sixth layer that works across many interviews: which planned questions get asked, how consistent interviewers are, and recordings used to train new interviewers and calibrate how they score.

How does interview intelligence capture an interview?

Capture is the part candidates notice, so it is worth understanding before anything else. There are three common methods.

• A bot participant. The tool joins the video call as an extra attendee and records from inside the meeting. It can attend calls the interviewer misses, but the candidate sees an unfamiliar name, and meeting platforms increasingly hold bots back. Microsoft's admin documentation for Teams says detected external bots are placed in the lobby by default until an organizer admits them (Microsoft Learn, as of October 2026).
• An app inside the meeting platform. Some products run inside one meeting client, so no extra participant appears on that platform.
• Desktop audio. An app on the interviewer's computer captures the microphone and the call's sound. Nothing joins the meeting, and calls routed through the computer, including phone screens, can be covered. See what is a bot-free AI note taker for how this works.

We make Interview Signal, which takes the third route: a desktop app that captures the interviewer's own computer audio, so no bot joins Zoom, Teams, Meet or a phone screen taken through the computer. It is one approach among several, and the buyer's guide compares the others.

What does interview intelligence look like in a real interview?

An invented example makes the difference from a plain transcript clear. A recruiter runs a 30-minute phone screen for a staff accountant. The intake meeting agreed three must-haves: month-end close ownership, reconciliation experience, and comfort with the company's ERP. Halfway through the call the candidate says:

"At my current company I run the close for two entities. I own the
bank recs and the intercompany entries, and last year I cut our close
from eight days to five by moving accruals to a checklist in NetSuite."

A transcript stops there. Interview intelligence files it against the plan:

Must-have: Month-end close ownership           Draft rating: Meets
Evidence: "I run the close for two entities"  [12:41]
          "cut our close from eight days to five" [12:52]
Follow-up to ask: Who reviews your entries before close is signed off?

Must-have: Reconciliation experience          Draft rating: Meets
Evidence: "I own the bank recs and the intercompany entries" [12:45]

Must-have: ERP                                Draft rating: Partly
Evidence: "a checklist in NetSuite" [12:55]  (the role uses a different ERP)
Next round: ask about learning a new ERP

The recruiter checks each quote against its timestamp, keeps the first two ratings, and adds the ERP question to the next-round plan. The judgment is still the recruiter's. What changed is that every claim on the scorecard now points to a line the candidate actually said, which is the core of evidence-based interview feedback.

How is interview intelligence different from similar tools?

Several kinds of product are described with similar words. They solve different problems and carry different risks.

The lines blur as vendors add features, so judge a product by what it produces, not by its category label. As of October 2026, BrightHire describes itself as an "interview intelligence platform" and HireLogic uses the same phrase, while Metaview calls itself an agentic recruiting platform whose note taker handles notes, scorecards and follow-ups alongside sourcing and screening agents. Our Metaview alternatives comparison covers how capture and pricing differ across products.

What are the common mistakes with interview intelligence?

• Buying it before you have structure. Interview intelligence sorts answers under your criteria. If the role has no agreed competencies and no scorecard, there is nothing to sort against, and you get a long summary instead. Start with an interview scorecard.
• Treating the summary as the evidence. A summary is a paraphrase. Ask for quotes with timestamps and spot-check them against the transcript before they reach a debrief or a client.
• Letting a draft score become the score. Drafts anchor people. Read the evidence first, then decide whether the rating holds.
• Switching on evaluative features by default. Sentiment, "fit" or automatic ranking change the legal picture. New York City's Local Law 144, for example, requires an automated employment decision tool to have had a bias audit within one year of its use, plus notices to candidates (NYC Department of Consumer and Worker Protection, as of October 2026). Our guide to Local Law 144 and note takers explains when it may apply.
• Skipping consent because nothing is visible. A tool without a bot still records or transcribes. Tell the candidate every time, with a line such as the one in our interview recording consent script.
• Measuring enthusiasm instead of outcomes. Track scorecard turnaround, how many ratings carry evidence, and must-ask question coverage before and after a pilot.

Is interview intelligence legal to use?

Generally yes, with two sets of rules to check. The first is recording consent. Some U.S. states require every participant to agree before a private call is recorded; Washington's statute, for example, makes recording a private telephone conversation unlawful without the consent of all participants, and treats an announcement made to everyone, and itself recorded, as consent (RCW 9.73.030, as of October 2026). The two-party consent states list covers the rest.

The second is the growing set of rules on AI in hiring, which mostly target tools that score, rank or screen candidates rather than tools that only transcribe. Which of them apply depends on where you hire and what the tool outputs. This is general information, not legal advice.

When should you use interview intelligence?

It earns its cost when at least one of these is true:

• Interviewers struggle to listen and take usable notes at the same time, so scorecards arrive late or thin.
• Decisions are made in debriefs where people argue from memory rather than from what the candidate said.
• Several interviewers share a loop and need to know which questions have already been covered.
• An agency has to turn a screen into a client submittal the same day.
• New interviewers need examples of good and weak interviews to learn from.

It is less useful when interviews are rare and informal, when no criteria exist yet, or when candidates in your market reliably refuse recording. In those cases a better scorecard, a structured interview and a note-taking habit will do more than software.