tl;dr

Develop a Gemini-Powered Voice Assistant for Gmail that transcribes voice inputs accurately, understands context, and composes high-quality, trustworthy email responses to enhance user productivity and satisfaction.

hy listen to me?

As a PM on Amazon’s Alexa, I’ve been working on the e-mail space for quite some time. In fact, we find that >12% of Alexa actions are now e-mails, up 200% from 3 years ago when I started.

Problem Statement

1. The current Gemini Advanced integration is text-only.
2. The voice transcription solution is literal-only. It is not AI enhanced.

The result is users can’t leverage the power of LLMs with their voice. They either type to use the LLM or use their voice for transcription.

Gmail users need a voice assistant that not just accurately transcribes, but understands context, recalls relevant info, and composes coherent responses users can trust.

Goals

User Goals:

• Quickly compose professional emails from natural voice
• Trust Gemini-generated content captures intent and tone
• Streamline workflows on mobile and desktop

Business Goals:

• Decrease Gmail churn >2%
• Boost Gmail user satisfaction 10%+
• Drive adoption of Gemini voice APIs >5%
• Establish Gemini as leader in AI voice communication

Non-Goals: