
Francis Bacon “Painting 1946”, 1946
artwork:
Pulkrae


Pulkrae is an easy to navigate AI-powered search platform that provides preview of artworks all across the web. The preview contains a link to a digital gallery that hosts the original, where one can download the piece in high resolution, if such is available.

Preview

Emphatize
There are always gonna be people interested in high art, whether it’s for educational purposes or for recreation. So having a platform for it is desirable for any art lover.
Define
With all the various online galleries providing public domain artwork in high resolution there is no platform that would connect all these digital libraries.
Ideate
A dedicated tool for online search of artworks is required. A search engine based around the Pinterest model featuring build-in AI search with an upload feature.
Competitor analysis
As there is a huge amount of galleries, no platform hosts the range of tools which define Pulkrae. Yet the project’s vision is heavily based on existing products features listed below.

Jstor
An academia centred platform with a huge image base as well as articles. A lot of it is limited to those users who sighed up through a school or a library accounts.

Artsper
European platform for online contemporary art sales with more than 130000 original artworks. The site features an AI powered search with a visual search feature.

Pinterest is a visual discovery platform where users can find, save, and curate images such as artwork with a Visit site button that redirects the user to the original image.
User personas
The two main categories of users are those who study art and those who like to experience art for the enjoyment derived from it.
Background & Goals
Personal Mission: To curate his own virtual gallery of favorite works and uncover new artists from around the world.
Exploration Goals:
Spend weekend afternoons browsing art platforms and virtual exhibitions.
Build thematic online collections (e.g., “Urban Landscapes,” “Surreal Dreams”).
Share favorite finds through social media or a personal blog.
Behaviors & Needs
Browsing Habits:
Visits platforms like Artsy, DeviantArt, Behance, and Instagram.
Uses keyword searches (“neo‑expressionism,” “generative art”).
Often follows links from museum and gallery websites to their online archives.
Source‑Gathering:
Saves high‑resolution images or links in browser bookmarks and a dedicated “Art Finds” folder.
Subscribes to RSS feeds or newsletters from favourite digital galleries and art blogs.
Tools & Platforms:
Browser with organized bookmark folders and the Pocket extension for saving pages.
Pain Points
Discovery Overload: Feels overwhelmed by the sheer volume of online content and struggles to filter signal from noise.
Organization: Finds it tedious to tag and categorize saved works consistently, leading to a cluttered archive.
Motivations & Values
Personal Curation: Takes pride in assembling thematic collections that reflect his evolving taste.
Digital Fluency: Enjoys experimenting with new online platforms and staying ahead of trends like AI‑generated art.
Opportunities for Support
Integrated Art‑Discovery Platform: A service that aggregates content from multiple art websites, applying smart filters based on his taste and past saves.
Automated Tagging: AI‑powered tools that suggest tags, styles, and metadata for each saved artwork to streamline organization.
License‑Aware Sharing: Clear indicators of usage rights and embedded share links that respect artist licensing.
Community Curation Workspace: A collaborative online space where Lukas and like‑minded explorers can co‑curate virtual exhibitions and comment on each other’s finds.
Name: Lukas Vogel
Age: 31 Location: Berlin, Germany
Background & Goals
Educational Focus: Aoife is passionate about 19th– and 20th‑century movements—particularly Impressionism, Cubism, and Abstract Expressionism.
Career Aspiration: She hopes to curate and eventually teach museum classes on modern art.
Behaviors & Needs
Study Habits:
Spends 3–4 hours daily in the library’s special‑collections reading room.
Keeps an “Inspiration Board” (physical corkboard) at home with postcard‑sized reproductions.
Source‑Hunting:
Requires high‑resolution images of original works (museum archives, digital collections).
References exhibition catalogs, critical essays, museum wall labels, and auction house provenance notes.
Tools & Platforms:
Uses JSTOR and Artstor for academic articles and hi‑res plates.
Follows museum websites (Tate, MoMA, Louvre) and subscribes to RSS feeds for new digitized archives.
Pain Points
Access Delays: Long wait times for inter‑loan requests of out‑of‑print catalogs.
Image Quality: Watermarked, low‑resolution reproductions that obscure brushstroke detail.
Fragmented Sources: Difficulty consolidating visual and textual references into one annotated workspace.
Opportunities for Support
Centralized Digital Repository: A platform that aggregates museum‑provided high‑res images, exhibition essays, and provenance records in one place.
Integrated Annotation Tool: Browser or desktop app that lets Aoife annotate images, link back to text sources, and export both in her preferred citation style.
Peer‑Sharing Feature: A collaborative board where classmates can upload and discuss key works, tags, and notes in real time.
Name: Aoife Brennan
Age: 22 Location: Dublin, Ireland
The personas are generated with a help of ChatGPT

Mind map
AI engines
These are the the tasks that the integrated API’s must be able to consistently accomplish:
Suggest similar artworks based on search history
Search similar artworks presented on the website
Generate descriptions of artworks while citing sources
Being able to determine whether the artwork is available for download in hig-res
Being able to generate tags based on the artwork
Scan the webpages for artworks and provide links to the original
The best choice is ChatGPT due to it’s ability to search the web. Which is crucial for generating descriptions with citied sources as well as analysing web pages for download links.
ChatGPT API
Content analysis
The two good choices for the visual search and image recognition would be VisionBrain or Google Cloud Vision with the former being more specialized and the later more generalised.
Google Cloud
Vision API
VisionBrain
Visual search

Information architecture

User flow
Wireframes
Home page unsigned
The background is a slideshow, circling between a few publicly available artworks, which creates a strong first impression upon the first visit utilizing the power of art. The big search bar in the middle gives a clear indication what the platform is about. The scroll is a set of non-personalised artworks as the user has not signed in yet.


The home page for a signed user contains a feed of artworks based on recent search history. There is also a display of on hover options on the artwork. The save button allows the user to save the artwork in a folder of choice or just save it if no folders were created. The High res icon near the website link indicates whether the artwork is available for download in high resolution. The icon is not displayed if the image available for download is less than 1080p.
Home page signed

The search results page features a row of chips based on generated tags in order to initiate another search with the selected categories.
Search results

The view artwork page contains: the artwork, the link to the original site with an indicator whether the artwork is available for download in high res, the save button with a dropdown menu to change the save folder, generated descriptions with cited sources and chips with search tags plus a quick search button.
View artwork

A simple folder system to organise artworks, with folder created and named by the user. It can be utilized as a mood board or a study tool.
Saved artworks
A rather shallow navigation allows for both the app and web to use the same navigation and UI elements.

Saved artworks

Search results

View artwork

Home page

Body | Descriptions
16 light
H5 | Prompts, dates
16 Regular
H4 | Chips, buttons
16 Medium
H3 | Author, save to button
20 Medium
H2 | Navbar desktop, boards
24 Medium
H1 | Artwork title
32 Bold
font family
Outfit
#FFFFFF
#EEEEEE
#222222
#414141
In order to ensure the quality of experience and to avoid distracting elements only achromatic colours are utilised. The spaciousness of the UI elements invokes an art gallery feel, while the artworks themselves are full screen width on mobile in order to enhance to viewing experience.
Style guide




Prototype


Pulkrae




Pulkrae is an easy to navigate AI-powered search platform that provides preview of artworks all across the web. The preview contains a link to a digital gallery that hosts the original, where one can download the piece in high resolution, if such is available.
Emphatize
There are always gonna be people interested in high art, whether it’s for educational purposes or for recreation. So having a platform for it is desirable for any art lover.
Define
With all the various online galleries providing public domain artwork in high resolution there is no platform that would connect all these digital libraries.
Ideate
A dedicated tool for online search of artworks is required. A search engine based around the Pinterest model featuring build-in AI search with an upload feature.
Competitor analysis
As there is a huge amount of galleries, no platform hosts the range of tools which define Pulkrae. Yet the project’s vision is heavily based on existing products features listed below.


Jstor
An academia centred platform with a huge image base as well as articles. A lot of it is limited to those users who sighed up through a school or a library accounts.


Artsper
European platform for online contemporary art sales with more than 130000 original artworks. The site features an AI powered search with a visual search feature.


Pinterest is a visual discovery platform where users can find, save, and curate images such as artwork with a Visit site button that redirects the user to the original image.
User personas
The two main categories of users are those who study art and those who like to experience art for the enjoyment derived from it.
Name: Aoife Brennan
Age: 22 Location: Dublin, Ireland
Background & Goals
Educational Focus: Aoife is passionate about 19th– and 20th‑century movements—particularly Impressionism, Cubism, and Abstract Expressionism.
Career Aspiration: She hopes to curate and eventually teach museum classes on modern art.
Behaviors & Needs
Study Habits:
Spends 3–4 hours daily in the library’s special‑collections reading room.
Keeps an “Inspiration Board” (physical corkboard) at home with postcard‑sized reproductions.
Source‑Hunting:
Requires high‑resolution images of original works (museum archives, digital collections).
References exhibition catalogs, critical essays, museum wall labels, and auction house provenance notes.
Tools & Platforms:
Uses JSTOR and Artstor for academic articles and hi‑res plates.
Follows museum websites (Tate, MoMA, Louvre) and subscribes to RSS feeds for new digitized archives.
Pain Points
Access Delays: Long wait times for inter‑loan requests of out‑of‑print catalogs.
Image Quality: Watermarked, low‑resolution reproductions that obscure brushstroke detail.
Fragmented Sources: Difficulty consolidating visual and textual references into one annotated workspace.
Opportunities for Support
Centralized Digital Repository: A platform that aggregates museum‑provided high‑res images, exhibition essays, and provenance records in one place.
Integrated Annotation Tool: Browser or desktop app that lets Aoife annotate images, link back to text sources, and export both in her preferred citation style.
Peer‑Sharing Feature: A collaborative board where classmates can upload and discuss key works, tags, and notes in real time.
Name: Lukas Vogel
Age: 31 Location: Berlin, Germany
Background & Goals
Personal Mission: To curate his own virtual gallery of favorite works and uncover new artists from around the world.
Exploration Goals:
Spend weekend afternoons browsing art platforms and virtual exhibitions.
Build thematic online collections (e.g., “Urban Landscapes,” “Surreal Dreams”).
Share favorite finds through social media or a personal blog.
Behaviors & Needs
Browsing Habits:
Visits platforms like Artsy, DeviantArt, Behance, and Instagram.
Uses keyword searches (“neo‑expressionism,” “generative art”).
Often follows links from museum and gallery websites to their online archives.
Source‑Gathering:
Saves high‑resolution images or links in browser bookmarks and a dedicated “Art Finds” folder.
Subscribes to RSS feeds or newsletters from favourite digital galleries and art blogs.
Tools & Platforms:
Browser with organized bookmark folders and the Pocket extension for saving pages.
Pain Points
Discovery Overload: Feels overwhelmed by the sheer volume of online content and struggles to filter signal from noise.
Organization: Finds it tedious to tag and categorize saved works consistently, leading to a cluttered archive.
Motivations & Values
Personal Curation: Takes pride in assembling thematic collections that reflect his evolving taste.
Digital Fluency: Enjoys experimenting with new online platforms and staying ahead of trends like AI‑generated art.
Opportunities for Support
Integrated Art‑Discovery Platform: A service that aggregates content from multiple art websites, applying smart filters based on his taste and past saves.
Automated Tagging: AI‑powered tools that suggest tags, styles, and metadata for each saved artwork to streamline organization.
License‑Aware Sharing: Clear indicators of usage rights and embedded share links that respect artist licensing.
Community Curation Workspace: A collaborative online space where Lukas and like‑minded explorers can co‑curate virtual exhibitions and comment on each other’s finds.
The personas are generated with a help of ChatGPT
Mind map


AI engines
These are the the tasks that the integrated API’s must be able to consistently accomplish:
Scan the webpages for artworks and provide links to the original
Being able to generate tags based on the artwork
Being able to determine whether the artwork is available for download in hig-res
Generate descriptions of artworks while citing sources
Search similar artworks presented on the website
Suggest similar artworks based on search history
Visual search
VisionBrain
Google Cloud
Vision API
The two good choices for the visual search and image recognition would be VisionBrain or Google Cloud Vision with the former being more specialized and the later more generalised.
Content analysis
ChatGPT API
The best choice is ChatGPT due to it’s ability to search the web. Which is crucial for generating descriptions with citied sources as well as analysing web pages for download links.
Information architecture


User flow


Wireframes
Home page
The home page for a signed user contains a feed of artworks based on recent search history.
Type or upload to search
12:45
View artwork
The view artwork page contains: the artwork, the link to the original site with an indicator whether the artwork is available for download in high res, the save button with a dropdown menu to change the save folder, generated descriptions with cited sources and chips with search tags plus a quick search button.


Search results
The search results page features a row of chips based on generated tags in order to initiate another search with the selected categories.


Saved artworks


A simple folder system to organise artworks, with folder created and named by the user. It can be utilized as a mood board or a study tool.
A rather shallow navigation allows for both the app and web to use the same navigation and UI elements.
Desktop










Style guide
In order to ensure the quality of experience and to avoid distracting elements only achromatic colours are utilised. The spaciousness of the UI elements invokes an art gallery feel, while the artworks themselves are full screen width on mobile in order to enhance to viewing experience.
#FFFFFF
#EEEEEE
#222222
#414141
font family
Outfit
Body
16 light
H5
16 Regular
H4
16 Medium
H3
20 Medium
H2
24 Medium
H1
32 Bold


Prototype









