Out & About

One of the things I struggle with the most when building a product, is trying to make the MVP as perfect as it can be. Perfect is the enemy of good, and while I know I should be getting the product out in front of people to get feedback, I often struggle with wanting to spend my time improving it before I even show it to anyone.

When I originally created the site, I did it through a combination of GitHub pages and GitHub actions - which are both free to me to use with my existing plan.

Given the nature of Artificial Intelligence and large language models, it tends to get things wrong, which means that as a human, I need to review it. For events gathering, it's easier for me to click a few buttons at the start of my morning than it is to go in and edit some JSON (which is what I was originally using as a data storage mechanism).

After a ton of hard work, I've converted it all to django site, running on a digital ocean droplet. It'll cost $5 dollars a month, but it'll be worth the ease of being able to quickly iterate, edit events, and do more things.

Fingers crossed!

My site serves no purpose unless people can see it. So with tons of events coming in every day through the AI / backend, it's time to start focusing on the front-end, and improving traffic coming to the site.

Right now, it's just social-media referred traffic, but my goal is to get it to the front page when you search "gay toronto events" or "lgbt toronto events" or some variation of that.

As you can see, the page only appears when explicitly using the site's domain in the search bar, so I've got a lot of work to do. Here's my general plan:

1. Run the site through the Lighthouse chrome extension, and fix any obvious answers.
2. Run some of the basic rendered HTML through ChatGPT and ask for any SEO suggestions.
3. Provision Google Search console, and fix any issues
4. Generate a site map and upload it to Google Search Console
5. Update the title to be simpler, remove the branding from the title.

Is there anything I'm missing?

Facebook's graph API is possibly the worst thing I've dealt with in years. Unfortunately, it's a necessary evil.

I want to meet users where they are - a daily story update to instagram.com/outandabout_toronto indicating the events that are happening today/tomorrow.

How It Works

Using a pipeline that took me a literal full-day to build, the following happens:

• Events for the day are gathered
• Events are displayed and rendered onto a page that's a lot more pretty than the normal site (colourful)
• A selenium docker image snags a screenshot of the page, with the appropriate dimensions for instagram
• The screenshot is uploaded to a public location
• The graph API authenticates using OAuth credentials
• The story gets posted

Now I've got a whole new problem on my hands, growing an instagram following (and not getting banned, which Facebook's AI pipeline loves to do).

I'm currently working on something I'm calling Project 6x6, where I try to build 6 AI projects/startups/websites in 6 months. The first, was liveabroad.fyi, and outandabout.fyi is the second.

Out & About is essentially an events aggregator for queer people to find events in their cities. I'm starting with Toronto, but the plan is to expand it out once I've got the bugs fixed.

The project is essentially a giant data pipeline making a ton of use of large language models in order to discover, filter, and display events.