Gerstler

Inspiration

With Google and Co., we can access any information quickly, but how about accessing our past knowledge and ideas? Do you know the problem of trying to remember where you found something that inspired you back then and could be useful now? With all the different apps and platforms, our knowledge is scattered, making rediscovery even harder.

We believe that a search engine for all our information sources is a powerful solution to this. Just enter a keyword and get instant results from past notes, documents, related media, links...

P.S.: The project name is inspired by Alexander Gerst, a German astronaut :)

What it does

The user is provided with a Web UI for entering a search query. The app connects to various services from the user and provides personalized live results.

How we built it

The search engine is powered by a Golang backend, that connects to various media resources like Google Drive, local file system, Notion and Gmail. We use the respective API services and aggregate the search results through an API endpoint. The frontend is built with React and sends search queries while typing to the backend. The file system crawler is based on Elasticsearch for high performance. All services are set up in a docker compose file to facilitate deployment.

Challenges we ran into

The Notion API is very limited, and we did not mange to fetch useful results with relevant content/context. A user-friendly setup for the integration of the services is difficult. Notion needs some manual steps for getting the access tokens and connecting a local file directory requires adjustment in the docker file. Performance is very critical, and the response time for Mail is not yet satisfactory.

Accomplishments that we're proud of

• live search results while typing
• concurrent processing of search from the various sources
• integration of 4 sources (Google Drive, local file system, Notion and Gmail)

What we learned

• the components and principles of a search engine

What's next for Gerstler

• improve search performance and relevance
• integrate more services like Readwise, Raindrop, Slack, Paperless (OCR document processing)

Built With

• docker
• golang
• react