The big problem I have with every knowledge base is that in order to retrieve knowledge you need to know that it is in there.
As a software developer my main tool for retrieving knowledge is Google. Although I document most of the things I learn in my Obsidian vault I regularly find myself googling the same things over and over simple because I can’t remember I’ve already solved them (and it was not worth remembering them via Spaced Repetition).
That’s why I created a search engine that connects to your obsidian vault via a plugin.
This completely changed my way of working. Whenever I find a solution for a problem I put it into my vault and the next time I search for it i immediately get the solution for it.
Here’s a short demo how it works at the moment:
SearchR Demo (Youtube)
Please let me know if you are interested in it (it will cost a few $s/month) so that I know if it’s worth publishing. This is just the first version - we can build this to whatever you want
This looks cool, but I would like to see more of the implementation before I allow a search engine to access to my vault.
This looks interesting, I’ve thought of building a tool like this before, but didn’t make my mind clear yet.
I noticed your demo was working in a website, are you planning to make this a SaaS service? I think many people use obsidian because they don’t want to put the data in the cloud, this might be a point to consider.
@jwl The search is done with minisearch. It’s quite similar to the Omnisearch plugin. Through express an secured API (CORS & auth key) is exposed that is used for the communication with the application
@reorx It’s a SaaS tool but all the data stays on your local machine - the application communicates with your vault over an obsidian plugin.
Absolutely love the idea! However I have a few questions. How are the search results order for the rest of the internet results? Are you using another well known search engine (Google/duckduckgo for example) and injecting results from obsidian in or is that also custom made?
(I would personaly probably prefer the results injection method as I regularly go through multiple search engines to find results)
Right now the “default” results come from Google. I even implemented filtering those for custom sites like Stackoverflow, Github,…
In addition I want to allow adding further personal search sources like Notion, Roam,… as soon as they have an API
Custom filtering is a very neat concept! I would probably still want the ability to look for pictures/videos/news, but having those extra filters (and heck even being able to define my own) is very cool!
Yeah sure - I can add the option to define custom dynamic links there that send you over to google/… with the searched query