autentik

autentik is a knowledge network where the focus is not on individual pages but on the connections between content, such as artists, albums, songs, places, products and topics. An interactive graph makes these connections visible and lets users navigate them.

The problem

Classic websites organise content into pages and menus. autentik was meant to work differently: the focus is on how things are connected. A Porsche 911 leads to commercials, photographers and songs, an artist to albums, concerts, genres and related artists.

The content comes from many sources and adds up to millions of entries. It had to be merged, linked and available quickly.

The solution

Graph navigation with D3.js: For every object, an interactive graph shows the connected content. Nodes are grouped by type and sorted by relevance, for example genres by votes, videos by views and albums by release date.
Configurable page types: The layout of a page is defined by page types made of modules, for example for artists, genres, playlists or products.
Embedded media: Maps, videos, Spotify players, timelines and record offers from Discogs.
Finder: Users answer a few questions, for the mountain bike finder about rider type, terrain and riding style, and see matching products.
AI chat: A chatbot with internet access answers questions about the content and can create playlists that are saved straight to the database.

Object page for the Porsche 911 F-Model with linked photos, ads and categories — Object page with linked content

Rolling Stone list of the 500 greatest songs with a Spotify player and album covers — Ranking with an embedded Spotify player

Page about Brooklyn with a satellite map, a city map and linked places — Places with embedded maps

Timeline of Porsche 911 models from 1963 to today — Timeline

Page about a Porsche commercial with an embedded video and linked categories — Embedded videos

Interactive D3.js graph with the connections of the Porsche 911 F-Model — Interactive graph of connections

Mountain bike finder with questions about rider type, terrain and riding style — Mountain bike finder

Detail view of a rider type in the finder with a graph of matching mountain bikes — Detail view with a graph of matching products

Finder results with rated mountain bikes — Finder results

Technical challenges

Merging data from many sources

The content comes from MusicBrainz, Wikipedia, Wikidata, the Cover Art Archive, Spotify, ListenBrainz and YouTube. Each source has its own formats and IDs. The MusicBrainz data is imported in full into the Neo4j graph database. Additional data such as texts, images, videos and Spotify links is only loaded when a page is visited, and then stored. This keeps the import lean while pages still get all the information.

Millions of entries for artists, recordings, albums and releases in one graph database
Ongoing development with regular new features, most recently a new design for artist pages, a player for ARD Sounds and personal playlists for users

More projects

Hostcraft
Hostcraft gives holiday hosts their own booking website with calendar sync to Airbnb and Booking.com. Built with Next.js, Payload CMS and Stripe.
Mart App
MART is an iOS and Android app that researchers use to run surveys and collect smartphone usage data. Used in a study at FU Berlin with 1,780 participants.
Crypto Tax Tool
The Crypto Tax Tool analyses CoinTracking exports: tax-free status with an adjustable holding period, holdings at current prices, gain and loss analysis and a transaction list. Runs entirely in the browser.

Contact Contact

Please call me or send me an e-mail for discussing your Project.
Depending on the service, an hourly rate between 60 and 80 € net is charged.

+49 1577 3125228

[email protected]