Great, how many days you took to launch TasteKit, for Botlytics you said it was done in 2 days.
Kindly share some insights on how to launch products 'fast'. Thanks
@tahaqadri Great question, Taha!
I recently went on a trip to Hawaii, and the idea came to me on the plane. The flight was 6 hours long, and I finished the majority of the core work then. Just this past weekend I spent more time to focus on the front end and API design, totaling around 8 more hours.
The reason I'm able to build things quickly is because I keep things simple and always have an MVP mentality. I determine what the absolute minimum is for the idea to launch, and iterate from there.
@jsngr Wow, that's cool, it reminds me of 'Limitless'.
Thanks for the reply and explanation.
Report
@jsngr@tahaqadri That is exceptionally amazing! Would love to learn more about how you even gained those skills. Seem to have to be a veritable Full Stack dev genius or at least fairly seasoned before attempting that.
Im an ideator practicing with Altucher's 'How to Become an Idea Machine' and would love to 'build my ideas' in a similar style.
Feel free to check out some of my ideas on my Twitter and see what you can do with any of that.
I will be checking out more of your stuff to figure out exactly how you do this haha.
TasteKit is a recommendation engine API. I built it because there were some occasions where I needed to implement Tinder-style liking and disliking into apps, and the complexity behind intelligently recommending things can get hard.
With TasteKit, you simply like and dislike things via the API, and are able to fetch recommendations. There's no need to dump all of your existing set of items. This is because TasteKit uses the Jaccard similarity coefficient. Recommendations are based entirely on what users like and dislike, we couldn't simply recommend you items immediately if you were to dump them all in the database. This is why items get added as users like and dislike things, and users with similar tastes are recommended things they haven't seen that have been liked prior.
For example, say you're using a movie app that recommends you movies: you like Toy Story and Finding Nemo. Another user likes Toy Story, Finding Nemo, and Finding Dory. Since you have similar tastes, Finding Dory will be recommended to you.
Report
@jsngr would the recommendations work if I only used the likes API and not the dislikes API? I'm thinking very simple ecommerce recommendations - if a user clicks on an item = like, show recommendations based on what other users clicked on.
@jsngr Epic idea. Love this and looks well executed, great docs.
Are you planning to charge $?
Also would love to know the stack behind it. How'd you build it? How long did it take from idea -> launch.
@mscccc Thanks!
Haven't thought about $ yet. Wanting to add more features and make it smarter before that happens.
It's built on Ruby on Rails, uses Postgres and Redis. I became interested in the Jaccard index when I was playing around with it after learning about it in school. The base of the code can mostly be found here: https://github.com/jordansinger/...
It was just a matter of making it so that apps could sign up and get their own API key. Like I mentioned in a comment earlier, the idea came to me to turn it into a service on the plane to Hawaii. Totaled about a days worth of work.
Nuage Email
Get After It
Automator for Figma
Get After It
Automator for Figma
Automator for Figma
Automator for Figma
PlanetScale Boost
Automator for Figma
PlanetScale Boost
Automator for Figma
Panda