How to Use Bridge AI: Key Information and Criteria Checks

Bridge Labs uses AI to analyze every candidate response against the standards you define. The quality of that analysis depends directly on two inputs you control when building each question: Key Information and Criteria Checks. This guide explains what each one does, how to write them well, and how the AI uses them to score and summarize responses.

​

Key Information

Key Information is a set of phrases, keywords, or concepts that a strong answer to this question should include. You add them as tags in the Key Information field inside the question builder. Examples for a question like “How would you handle a conflict between two team members?”:

• Active listening
• Neutral mediator
• Root cause
• Documentation
• Follow-up

When a candidate answers, Bridge AI scans the transcript for these terms and concepts. Matches are flagged in the analysis and contribute to the overall score. Candidates whose answers hit more of your key terms score higher.

​

Tips for writing good Key Information tags

• Be specific. “Budget management” is more useful than “finance.”
• Use the language your top performers actually use, not just formal HR terminology.
• Aim for 3 to 6 tags per question. Too few gives the AI insufficient signal; too many dilutes what matters.
• For technical roles, include specific tools, methodologies, or frameworks the candidate should reference.
• For soft-skill questions, include behavioral indicators: “took ownership,” “asked clarifying questions,” “proposed a solution.”

​

Criteria Checks

Criteria Checks are a checklist of specific behaviors or content elements that define what an excellent answer to this question looks like. While Key Information looks for keywords in the transcript, Criteria Checks evaluate the structure and completeness of the candidate’s response at a deeper level.

​

Generating Criteria Checks with AI

Rather than writing criteria from scratch, click Generate Criteria Checks. Bridge AI reads your question text and Key Information tags and generates a set of criteria items automatically. The generated criteria cover the most important dimensions of a strong answer. Review them and:

• Check the box next to each criterion you want to keep active
• Edit any criterion by clicking on it to adjust the wording
• Add your own criteria by typing in the “Add another criteria check” field at the bottom
• Remove any criterion that does not apply by unchecking or deleting it

​

How to write strong Criteria Checks manually

Each criterion should describe a specific, observable element of a strong response. Write them as statements the AI can verify against the transcript:

​

How Bridge AI Uses These Inputs to Score Responses

When a candidate submits their interview, Bridge AI processes their video responses and runs the following analysis for each question:

1. Transcription: The candidate’s spoken response is converted to a full text transcript.
2. Key Information matching: The transcript is scanned for the keywords and phrases you defined. Each match is noted.
3. Criteria evaluation: Each active Criteria Check is assessed against the full transcript to determine whether the candidate’s response satisfies it.
4. Scoring: A question-level score is calculated based on keyword coverage and criteria satisfaction. Scores across all questions are averaged into an overall candidate score.
5. Summary generation: Bridge AI generates a written summary of each response that highlights strengths, gaps, and notable moments relative to your criteria.

​

Using Bridge AI for Mock Interviews and Practice

Criteria Checks and Key Information are equally powerful in a practice and coaching context. Coaches can build questions that define exactly what a polished, interview-ready answer looks like, then share the scored results with clients as objective feedback. Instead of subjective coaching notes, the client gets a structured analysis of which criteria their answer met and which it missed. Schools can align criteria checks with rubric standards, ensuring every student’s practice response is evaluated against the same consistent benchmark regardless of who reviews it.