More relevant content recommendations, using AI

Would you recommend this product?
No reviews yet
Thanks @edrex for putting us up on Product Hunt! At Bibblio we're helping solve the problem of discovering the best content in a world of too much choice. To do this we've launched a SaaS product that harnesses AI to enable publishers and content platforms to provide users with smarter, more relevant recommendations. Our API uses natural language processing to quickly understand large volumes of content and identify key subjects and concepts. It then uses those subjects and concept to identify the most relevant recommendations for users as they interact with content on a platform. Unlike current recommender engines, like Taboola and Outbrain, it's optimised for relevance and diversity as well as popularity, so no filter bubbles and no clickbait! We really appreciate the interest and feedback of the Product Hunt community, so we're giving away £500 of API credit to the first 200 sign-ups from Product Hunt - - check it out!
Upvote (21)Share
@madsholmen @edrex Hey Mads, this sounds super interesting to me. I'm the founder of KYA ( and we're a new kind of analytics platform for digital publishers/content creators. One of the additional features we offer is a super personalized content recommendation widget our customers can embed on their site to recirculate their content based on a reader's previous engagement on their site. I'd love to chat with you about Bibblio in more detail, perhaps there's a way we can work together or at least bounce ideas off each other. :)
@jeffweisbein @edrex Happy to chat further, sounds interesting.
@madsholmen @edrex Awesome! Is there a best email or way to get in touch? You can email me at
@jeffweisbein @edrex @getkya Great, I'll drop you a line.
Bibblio is structured serendipity, a tool to map knowledge. It allows you to truly discover rather than be served up what you already know. And their APIs are so damn clean.
Thank you @r_nash!
Love the look of Bibblio - great job everyone involved. It seems to me this fits into a wider trend of personalisation. How do you provide recommendations - and are they 'personalised'?
@edrex_ That's a really interesting question. Personalisation approaches are proving very popular, but currently they definitely have their limits - Demis Hassabis at DeepMind recently stated that “Personalisation doesn’t work very well. It currently sums up to averaging the crowd as opposed to adapting to the human individual”. "Averaging the crowd" is what tends to push less relevant but more popular content to the top of recommended and related content lists. We actually wrote a blog post on this recently: "Popularity vs. Diversity" - What we do aims to "adapt to the human individual" by finding what's most relevant to people as they are browsing content. By 'understanding' subjects and concepts in the content Bibblio can learn more about the key themes that interest users and look for other content which contains related themes (but in a much more sophisticated way than simple keyword matching). This means people are more likely to get recommendations that will interest and engage them, whilst ensuring they are exposed to diverse content.
Looks great! Wanted to create an account but got this: "Password must be at least 8 characters long, and contain both upper and lowercase letters, a digit and one special character of -+=><_$#.:;!?@&*()~][}{|." It comes from a good intention but I hate these special characters! (And I guess I am not the only one)
@jhondel Jonathan, I fully understand where you are coming from. I actually had the conversation with our dev team and they highlighted that strong security is prudent in this instance (and you won't need it very often). Much appreciate your input to the conversation though!
@jhondel Have to agree with Jonathan. Would it be better to make a longer length requirement than include special characters?
@jeffrey_wyman @jhondel Great feedback guys. A ticket has been created...
This looks super interesting. Are there more options planned for the discovery endpoint? Searching all my documents for relevance is definitely interesting, but i'd love to be able to filter the documents being recommended on your side rather than mine. For instance: Being able to tell the API to limit your recommendation of a certain document to only other documents created by the same publisher. (e.g. only recommend me articles from within the documents that have the publisher set as NYTimes)
@daniel_schwartz Thanks Daniel! We are definitely planning on adding to discovery. We’ll be rolling out additional endpoints to support more user-driven exploration and search, as well as broader exploration around particular topics and themes. We will also be adding filtering options to the recommendation endpoint. We’re looking into filtering by existing fields on content items, as well as custom faceting on data points of your own. You’ll also be able to query across multiple catalogues — your own as well as high-quality, curated open content. Drop us an email and we’ll let you know when we roll out these features! We want to make them as clean as possible on the api, and might ask you to beta test them if you’re up for it. Thanks again for the feedback. It’s great to know that this would be helpful to you.