Skip to content

Contextual Tips

Get key points and confidence scores with your answers.

Overview

Contextual Tips provides structured answer output with:

  • Key Points: Bullet-point summary of main ideas
  • Confidence Score: How confident the LLM is (0-100)

This helps you quickly understand what to cover before reading the full answer.


Enabling Contextual Tips

  1. Open Settings (gear icon)
  2. Scroll to Advanced Settings
  3. Toggle Contextual Tips on

How It Works

Without Contextual Tips

The LLM generates a plain text answer:

I've been working with React for about5 years. Started with class components 
and then moved to hooks when they became mainstream. Most recently, I led a 
team building a real-time dashboard that handles 100K+ concurrent users using 
WebSockets and optimized re-renders...

With Contextual Tips

The LLM generates structured output:

KEY POINTS:
- 5 years of React experience
- Started with class components, transitioned to hooks
- Led team on real-time dashboard project
- Handles 100K+ concurrent users

CONFIDENCE: 85

ANSWER:
I've been working with React for about 5 years. Started with class components 
and then moved to hooks when they became mainstream...

Displayed in UI

Intervu parses this and displays:

  1. Key Points as bullet points above the answer
  2. Confidence as a visual indicator
  3. Answer as the main content

Using Key Points

During the Interview

Key points give you a quick overview:

  1. Glance at the bullet points
  2. Structure your response around them
  3. Expand with your own words and examples

Example Flow

Key Points Show:
- 5 years React experience
- Led team on dashboard project
- 100K+ concurrent users

You Say:
"I have five years of experience with React. One project I'm particularly proud 
of is a real-time dashboard where I led a team of five engineers. We were able 
to handle over 100,000 concurrent users through careful optimization..."

Understanding Confidence

What Confidence Means

Confidence indicates how certain the LLM is about the answer:

ScoreMeaning
90-100Very confident — answer is likely accurate
70-89Confident — good answer with minor gaps
50-69Moderate — answermay need verification
0-49Low confidence — verify before using

When Confidence is High

  • Question is clearly related to your resume
  • Standard interview questions
  • Technical questions in your domain

When Confidence is Low

  • Question outside your experience
  • Vague or ambiguous questions
  • Topics not covered in your resume

Interpreting Low Confidence

Don't dismiss low-confidence answers entirely:

  1. Use as a starting point
  2. Add your own experience
  3. Clarify with the interviewer if needed
  4. Mark as "I'm not as familiar with that"

Customizing the Prompt

The structured output is injected into the system prompt:

Structure your response as:

KEY POINTS:
- [key point 1]
- [key point 2]
- [key point 3]

CONFIDENCE: [0-100]

ANSWER:
[Your detailed answer here]

You can modify this in your custom system prompt if needed.


Performance Considerations

Token Usage

Contextual Tips increases token usage:

ModeApproximate Tokens
Without Tips200-500 per answer
With Tips250-600 per answer

The increase is minimal for most use cases.

Response Time

No significant impact on response time.


Best Practices

When to Use Contextual Tips

Good for:

  • Quick-glance interviews
  • Need to understand key points fast
  • Want confidence indication

Not needed for:

  • Reading full answers anyway
  • Very short answers
  • Rapid-fire questions

Combining with Other Features

Contextual Tips works well with:

  • Answer Ratings — Rate based on key point accuracy
  • Advanced Mode — Tips for extracted questions
  • Queue Mode — See key points for multiple answers

Example Output

Technical Question

Question: "What's your experience with TypeScript?"

Key Points:

  • 4 years TypeScript experience
  • Migrated JavaScript codebases to TypeScript
  • Strong type safety practices
  • Experience with generics and utility types

Confidence: 92

Answer: I have four years of experience with TypeScript, starting with simple type annotations and progressing to advanced features. I've migrated several JavaScript codebases to TypeScript, which improved our code quality significantly...


Behavioral Question

Question: "Tell me about a time you had to learn a new technology quickly."

Key Points:

  • Learned Docker in one week for deployment project
  • Had to containerize 5 microservices
  • Used official documentation and community resources
  • Successfully deployed before deadline

Confidence: 78

Answer: At my previous company, we needed to containerize our microservices for deployment. I had a week to learn Docker from scratch. I started with the official documentation, broke down the learning into...


Troubleshooting

Key Points Not Showing

Causes:

  • Contextual Tips is disabled
  • LLM isn't following the format

Solutions:

  • Verify the setting is enabled
  • Try regenerating the answer
  • Check if model supports structured output

Confidence Always Low/High

Causes:

  • Model isn't calibrated for confidence
  • Prompt needs adjustment

Solutions:

  • Try a different model
  • Adjust system prompt to emphasize accurate confidence scoring

Key Points Incomplete

Causes:

  • Answer was cut off mid-generation
  • Model hallucinated key points

Solutions:

  • Regenerate the answer
  • Rate down to indicate poor extraction

Next Steps

Made with ❤️by Aldrick Bonaobra