Benjamin Fletcher

Chat Essential - Turn a URL into a powerful AI assistant in 1 click

Enter a URL. Our platform will scan your website and create a virtual assistant that can talk about the content in seconds. Powered by OpenAI's GPT-3, Facebook AI's BART, and our own models.

Add a comment

Replies

Best
Olivier ATNORM
very cool and very simple to use
Benjamin Fletcher
@olivier_atnorm thanks Olivier for checking it out Olivier
Daniel O'Brien
Feels like it will have broad appeal in helping businesses answer questions quickly and accurately
Benjamin Fletcher
@dobrienco Thanks for taking the time to check us out, Dan
mike_rod
Super interesting. Congrats on the launch!
Benjamin Fletcher
@mike_rod Thanks Mike!
Yash
Hello, Can we export the trained data to use it in our other apps ?
Benjamin Fletcher
@yash094 possibly, can you give an example app?
Yash
@benjamin_fletcher Like using it as a chatbot in discord?
Benjamin Fletcher
@yash094 yes, we have a set of developer APIs that can be used for lower level integrations
Stephanie Ludwig
Thanks for all the comments, everyone! We're so excited to have finally launched. Our goal was to create the simplest onboarding for a very complex interaction model - conversation. We're really proud of how we found a solution that combined ease-of-use with a high level of accuracy.
Jeff Shenk
This is remarkably good! Great job.
Benjamin Fletcher
@jashenk thanks for checking it out, Jeff
Benjamin Fletcher
Hey everyone, thanks for taking the time to check out our product. When you enter a URL, we look for a robots.txt or sitemap.xml. If the website does not have one of these, our demo will not work :( If we're able to find a sitemap, we automatically extract the page content from the website (for this demo, with an imposed limit of up to 100 main site pages and 100 blog articles) and use it to power a custom virtual assistant that can talk about the content. Each time someone asks the virtual assistant a question, we run the input through a blend of our own technology, Open AI's GPT-3, and Facebook AI's BART MNLI models to auto-generate an answer. If we can extract clean content and there is enough of it, we've seen the auto-generated answers effectively respond to up to 89% of inquiries. The demo is most time optimized for marketing websites. We've noticed that it does not do so well with websites with huge product catalogs or lots of multimedia. But that doesn't mean our platform can't work for these types of use cases. We have manually uploaded a wide variety of content into the platform for customers, including scanned PDFs, Word documents, and internal company websites. If this is something you're interested in, feel free to reach out.
Arman Zand
Very cool!
Benjamin Fletcher
@arman_zand thanks for the feedback, Arman
Matt O'Brien
Love this! Worked perfectly for me
Benjamin Fletcher
@matt_o_brien5 thanks Matt!
Stephanie Ludwig
Bill Sedgwick
This is a great concept and a great product. Well done, all!
Benjamin Fletcher
@bill_sedgwick thanks for checking it out, Bill!
123
Next
Last