OutAndAbout.FYI

AI Projects

A website that lists all the current LGBTQ+ happenings and events in your city.

OutAndAbout.FYI

As a part of project 6x6, where I build six artificial intelligence projects in 6 months, I built outandabout.fyi.

What was it?

Simply, OutAndAbout was a website that lists the latest LGBTQ+ events happening in a city. It uses AI to find, and parse events, and presents them all in one place.


Why is it needed?

When moving to a new city, finding queer friends can be tough. The goal of OutAndAbout is to provide a list of all events happening in the community, so that those looking for connection can find the events that they'd like to go to, all in one place. From parties, to D&D, to drag.

How it works

The site is a simple front-end application supported by an advanced data pipeline, which worked like this:

First, find the sources

First, a web-crawler and query manager sourced queer organizations in given cities (Toronto, Vancouver, San Francisco). Once a potential source was identified, such as a sports group or community centre, its data was normalized and processed with the help of fine-tuned large language models (LLMs) and added to the database.

Next, get the events

Once a potential source was identified, it was monitored for any potential events. Any changes in the content of the website, or in the case of a social media account (like instagram), any new posts to the account were run through a separately fine-tuned large language model to determine:

  • If the post or content contained an event
  • If that event was a one-off or recurring event
  • What type of event was this, and was it relevant to the site
  • The details of the event, such as time, place, host, cost
  • Metadata about the event, such as images

Once the event was added to the database, a separate process researched more information about it to generate a comprehensive human description (where possible), and a categorizer process categorized the event using a standard large-language-model API (ChatGPT's gpt-4 model) into discrete categories for the site.

This pipeline takes events from various sources, and curates them with the help of artificial intelligence. It creates events from unstructured data, and categorizes them for what's best for the community.

The site is designed to run fully autonomously, but of course, like all things AI, it does require a bit of babysitting now and again.


Project Shutdown

Ultimately, the project wasn't seeing the traffic it needed to keep it running. It was however a great experiment into AI parsing of events from non-typical sources, like instagram stories.

Here are some photos from when the project was running.

A list of Sports teams

A list of LGBTQ+ sports relevant to the city as identified through the crawler.


An individual LGBTQ+ sports organization in Toronto.


An event parsed by the pipeline from an instagram post.


The About Us section of the site.


An instagram post of the day's events, scheduled to post automatically to the Out&About Toronto instagram.

Interested?

Check out the project for yourself!

Let's build something big together

I build creative marketing experiences and projects with technology.