Welcome new users! We’ve nearly doubled our users in the past 48 hours, and you’re getting this email about VI improvements so you can get the most out of what we’re building and we can keep ourselves accountable for shipping every day.
Have feedback?
We would love to talk to you. Setup a time to chat with Jonathan and Elle: https://calendly.com/d/x43-5vf-r8p/vi-customer-interview
VI is like ChatGPT, but VI can read the internet… and your email!
Yup! VI can search, summarize, and answer questions based on your email. We think it’s pretty cool because email has so much context about our work, contacts, and style of communication that can train VI to provide even more rich and personalized responses.
Here’s a simple example of what VI can do with email data as part of its corpus:
Want to try VI’s Gmail Integration?
We’re manually granting access for VI to read your Gmail account, and we’d love for you to try it out.
Fill out this form to request Gmail access
Right now we are manually provisioning and have ~80 spots left. Gmail access will be generally available after we will go through the Google approval process, but for now we are keeping it scrappy.
We have the capability to read the content of emails in the database and they are not encrypted. While we don’t plan to do this, it might be necessary at some point for debugging. For now if you aren’t comfortable with this, we don’t recommend using the Gmail integration until it is a more secure and mature feature.
From You!
Shaina Rozen of Sidetone showed us how VI was able to write a concise tweet about a webpage with no additional context:
Compare this to ChatGPT’s response to the same question, which looks good at first glance, but on closer inspection reveals that the content is inferred from the URL, not actually based on the content of the page. ChatGPT also uses an incorrect Twitter handle which was again inferred from the prompt.
What’s New
What’s Next
What else would be cool? What else do you wish VI could do?