Skip to content
On this page

Developer Blog

Directus AI Hackathon: Project List & Winners

Published September 1st, 2023

Written By
Kevin Lewis
Kevin Lewis
Director, Developer Relations

Winner Announced!

Check below as we've updated this post with our two winners for the Directus AI Hackathon.

Throughout August, we ran our very first Directus hackathon around the theme of AI. We challenged our community to build extensions that utilize AI in fun or interesting ways. We've had four valid submissions, and now it's your turn to get involved.

The core team will vote for our favorite project judging projects based the following criteria with no specific weighting:

  1. Creative: Is the idea original and innovative?
  2. Technical: How effectively do you integrate AI? Is the project well-executed? Does it run?
  3. Relevance: Do you solve a real problem faced by people?
  4. Presentation: Have you taken time to add polish to your project?

Community Vote

One of our prizes is decided by the community. From the submissions below, the repository with the most GitHub Stars before September 10 at 23:59 UTC will win this category. No stars given or removed after the end of the judging period will be accepted.

The Winners

Media AI Bundle - Community Winner

A gif showing a file uploaded, adn the description for the image being auto-populated.

This bundle contains two new operation extensions to be used within our Flows automation builder.

The first operation takes an image as an input and then describes what's in the image using or Amazon Rekognition. The operation results both a description that can be used as alt text, and the full raw API response.

The second operation uses Amazon Rekognition to extract text from an image, which can then be used in futher operations.

Check out the repository: Arood/directus-extension-media-ai-bundle.

Directus Copilot - Core Team Winner

A dashboard with a ChatGPT-like panel. Questions asked are 'how many customers do I have' and 'How may units do I have for product SKU WT'

This is a bundle of extensions - the first is a panel extension allows users to ask contextual questions about their data within Insights dashboards.

The second extension is a custom endpoint, required by the panel to work. The endpoint actually conducts requests to OpenAI.

Check out the repository: programmarchy/directus-extension-copilot.

The Projects

OpenAI Automatic Translation

A screenshow of a flow triggered on item dot create and the use of this custom operation.

This operation extension automatically translates content in collection items using the OpenAI API. In the README example, a new collection item is created in English, and within minutes, a further 7 languages are available in the translations interface.

Check out the repository: timio23/directus-operation-auto-translate.

Machine Learning Operations

The first panel describes training control settings with a button to 'Train Now'. The second is a prediction control input. And finally, an output which is populated when the user clicks predict.

This project is a set of extensions allowing developers to train, test and use machine learning model inside their existing project. Contained within this bundle is an operation, an endpoint, a hook, and two interfaces for use in the Directus Editor.

Check out the repository: karamokoisrael/directus-hackathon-submission.

Tell us what is missing

How helpful was this article?