Glimpse

The modern AI tool to ensure founders never pitch blind again. Glimpse analyzes both presentation delivery – scoring tone, pace, and confidence – and your pitch deck for narrative flow and clarity.

Comment

Inspiration

There is an endless amount of brilliant ideas in the world, but many often don't see daylight because of a fumbled pitch. The feedback loop for founders is either non-existent, outrageously expensive, or comes too late. To solve this, we built a tool that acts as a dedicated pitch coach, using AI to provide founders the confidence they need to tell their story effectively and secure the funding they need.

What it does

Glimpse is a comprehensive suite of tools designed to perfect any founder's dreams of pitching.

  • Live Practice: Glimpse's main feature is the ability to rehearse your pitch as if you were standing in front of a venture capitalist. Imagine if this was the real deal. The day you were pitching to the VC. Only difference is you get infinite tries, it's not a shot in the dark. Glimpse acts as a real VC responding to every thing you say as you are pitching.

  • Pitch Analysis: After your live practice sessions, our AI transcribes your speech and provides a detailed report breaking down your performance. It also provides insight to what the VC was thinking as you were pitching.

  • Pitch Deck Analyzer (Future Feature): Upload slide decks and our AI will assess it for design consistency, text density, logical flow, and the actual content of the slides.

  • Glimpse Report: For every analysis, you will receive an actionable report containing an overall score, highlights (strengths of your pitch), and also checklist of what to do next.

How we built it

  • Frontend: Built with React and Next.js, shadcn, tailwindcss, and LiveKit components
  • Backend: LiveKit, python, fastapi
  • AI: Whisper, DeepGram, Websockets, Gemini API
  • Database: MongoDB

Challenges we ran into

Our biggest hurdle was live feedback — making sure the AI could reliably listen and speak back in real-time. Getting audio sync across websockets while maintaining conversation context took some heavy debugging and iteration.

Accomplishments that we're proud of

  • Built a full-stack live AI pitch coach in < 36 hours
  • Achieved real-time voice input and LLM-driven responses and analysis
  • Created a user experience that genuinely feels like “practicing with a VC”

What we learned

  • How to blend multiple AI tools (speech-to-text, LLMs, synthesis) into one seamless feedback loop
  • Managing low-latency audio streaming in production-level LiveKit integrations
  • Prompt engineering for dynamic, role-based feedback

What's next for Glimpse

  • Voice sentiment analysis for deeper emotional feedback
  • VC firm personas – simulate pitching to YC, a16z, Sequoia, etc.
  • Pitch benchmarking – compare your pitch performance with peers
  • Multi-modal review – integrating body language and facial cues (via webcam)

Built With

Share this project:

Updates