A Binary Built for Two

 To recap from the previous post:

Me: Ok, computer, I want to start an online travel agency. Help me get started with the basics.

ChatGPT: Well, I'm glad you've asked me to help out again for this second blog post.  I'd like to continue to showcase my abilities to your readers.  Here are 8 pillars for creating an online travel agency...we'll address the Technology and Website section in more depth today.

*Author's Note: ChatGPT didn't say this during our recap, but wouldn't it be delightful if it did?

What ChatGPT actually said regarding Technology and Website Development:

• Choose a domain name: Select a memorable and relevant domain name for your website.
• Develop your website: Hire a web developer or use website builders like WordPress, Wix, or Shopify to create a user-friendly and visually appealing website.
• Integrate booking tools: Partner with travel suppliers or use third-party booking engines to offer a wide range of travel options to your customers.
• Ensure mobile compatibility: Optimize your website for mobile devices to accommodate users who prefer to book travel on their smartphones or tablets.

I suppose that, when casting the broad net that I did, I should expect the advice across all 8 pillars to be necessarily vague.  But, specifically regarding the technology section, the above advice could be applied to any Silicon Valley idea pitched in the last 15+ years and offers no particular thread to pull to get started.  If I paid for this advice, I would be upset.

Let me start by giving the suggestions the same treatment I did with the other 7 sections before delving into them a bit more afterward:

• Choose a domain name [I agree.  Choosing to refer to your site only by its IP address can get a bit tedious.  I suppose Coastal Chicago LLC is memorable to me.  The jury's still out on its relevance.  The name that I've selected for our budding OTA probably isn't all that memorable, but it is partially relevant.  I'll save the big unveiling for the Name Reveal Blog Post in the future.]
• Develop your website [This is what I'm expecting you to do, you highly distributed vast neural network, you!]
• Integrate booking tools [For those of you not familiar with the travel industry, there are 3rd parties that make this easier for smaller operations - or so we hope when we connect to them later.  But at anything greater than a micro scale, companies dedicate whole teams to this functionality.  Those teams are often split further by the product offered (air, hotel, car, etc.) and it's not uncommon to be segmented further within those products.]
• Ensure mobile compatibility [This is another suggestion that seems to hint that it's something that can be started on a rainy Wednesday afternoon and wrapped up by the weekend.  Though mobile teams are a smaller subset of the greater tech organization, they're integral to the overall strategy of the site and touch much of the functionality that other engineering teams deal with as well.]

Let's dive into (or double-click, as the corporate set liked to say in the early 2020s) the bullet points a bit more.

Choose a domain name

Setting aside the actual work here - finding a memorable, quirky, or relevant name - the process is actually pretty straightforward.  Up to this point, I've used Google Domains.  However, Google recently sold off this business to Squarespace.  I'm assuming (well, hoping really) that not much changes after all of the processes are moved over to Squarespace.  If the assumption holds, then I can tell you that I was able to score a few domains for no more than $13/year and that the Google domains maintenance page is straightforward.  Some of the functionality is geared toward people with more technical expertise (specifically when adding DNS records), but it's apparent that they're trying very hard to allow anyone to stand up a website without needing outside help.

As for names, I can't say that I'm a branding expert, but I lean toward "the get anything out there and then iterate" school of thought.  Starting something in the software world is cheap, so the cost of initial mistakes is low.  Once your brand has some traction, you'll have to adjust your risk calculation, but right now, if I changed "Chicago Bot Dog" to "The No Ketchup Club" I imagine there would be little blowback. 

And if, you're wondering why Chicago Bot Dog - well, most of the domains I own are Chicago-themed because I live here.  I knew I was going to set up a blog centered around technology and wanted to see if I could create some wordplay on Chicago icons.  Chicago-style hot dogs are fairly well-known, and it didn't take me too long to pull that from Chicago symbols and apply tech-adjacent punnery to it.  From there, I was able to choose the color scheme fairly easily - yellow for mustard and red for ketchup (and, yes, I realize that I just lost any Chicago-based business with that hot take.  Good thing I have no strong ties to the city!  For what it's worth, I would never put ketchup on a Chicago-style dog, but I will reserve the right to put ketchup on other hot dogs - especially picnic fare).

Develop your website

And now we've arrived at the heart of our project - to have generative AI tools write as much software for us as possible.   I think it's worth mentioning my expected conclusion upfront - there is no way at this point in time that a computer can sufficiently substitute for a software developer/software engineer (or for any other profession) at this juncture.  

I'm going to save my deeper thoughts on this for the next post, because I think it's important to address the state of the industry in full, and my thoughts are rather lengthy.  For now, I want to concentrate on the verbatim suggestions ChatGPT provided.

Let's start with the website builders.  Wix and Shopify are specifically geared toward non-technologists who want to open an online storefront.  Shopify focuses on the storefront message more succinctly than Wix does, but they're similar enough within the context we're concerned about here that we can treat them as equals.  

Both of them have easy drag-and-drop tools for site creation, inventory management, and payment.  This is wonderful if you want to sell some sort of product directly to a client.  They handle all of the infrastructure needed to maintain a website, so you don't need to worry about the site going down or (most of) the security and performance concerns that accompany running a site.  That's why you're paying a premium - ease of use and peace of mind.  

To some extent, they're also banking on their users' naïvety with technology to upsell certain add-ons.  They're not wrong, per se, because people often struggle with website building, even when the solutions are labeled as "no-code."  That's also why there are side industries either training people to use these tools for their businesses or doing the work for them directly (I'm helping a friend maintain a non-profit site on Wix due to this skills gap).   

This is where having someone who's technical consult on even the extreme basics of your business idea can be useful.  The basic solutions for Wix and Shopify start at $29/month or $348/year.  Now, if you ask anyone with a rudimentary background in software engineering if these types of sites can support an online travel agency, they should be able to tell you "no" very quickly.  

Why?  An OTA isn't a typical storefront, because you're not selling the service or product directly.  You're acting as a broker, adding your own differentiating experience, and taking your cut for that differentiating experience.  If things work out well, you have to worry about the scalability of the system in a manner the e-commerce platforms aren't set up to handle for their clients.

So, assuming you pay your consultant $100/hour and have a nice leisurely chat about your options at a high level, you'll only be out $100 with a better idea of where to focus.   If you're taken aback by the argument that saving yourself $248 a year when your total costs will begin to soar into the thousands (or millions!) and is just pennies on the dollar, that's valid.  But if you didn't know that you shouldn't use an e-commerce site to begin with, you're likely to waste waaaaay more than $248 and need a consultant anyway.

So, we've now encountered the first demonstrable instance where you can use GenerativeAI as a guide but, if you were to follow it verbatim, you'd get burned.  It's an obvious case, but it's also a good place to point out that, while you may trust your robot, always count the punch cards.

Wordpress, as it's mentioned in this context (...WordPress, Wix, or Shopify to create a user-friendly...), would suffer from the same flaws as the other builders, because it's fair to assume that ChatGPT is referring to Wordpress.com (the e-commerce offering) here vs. Wordpress.org (the open source build-your-own site).  The latter would necessitate hiring someone with technical acumen to stand up the site, configure, and maintain it.  I think you can run a start-up travel site using Wordpress.org, but there's no getting around some sort of development help for a site that isn't a mere triviality.

It appears then, that we've come to the conclusion that we'll need some sort of developer for our site.  Whether or not that role can be served by ChatGPT or needs someone in meatspace - well, that's what we're here to find out, isn't it?  Realistically, competing in the travel market will require teams of developers in addition to other supporting functions, but probably fewer than many travel technology companies currently think they need (we'll get into Conway's Law and the pains of empire-building later, but for now expect that an online travel site isn't built to be a one-person venture).

Integrate booking tools

This landscape has changed a bit since I first started playing around with a similar proof of concept about 10 years ago.  Providers are much stingier in allowing too much unfettered developer-level access, ostensibly because providing this access can cause a drain on their systems, which would lead to mounting costs without getting anything in return.  Among the contenders we can use who likely won't set us back too much for our initial expedition, Amadeus seems to provide pretty robust access to its test environment.  

For those not familiar with Amadeus, it's a well-known, European-based Global Distribution System (GDS).  A GDS is a clearing house for airlines and other travel options that just about everyone has to go through to make an actual travel booking.  

Back before the advent of the internet, travel agents - even your mom-and-pop variety - would log in to these systems or call their representatives directly to make a booking on your behalf.  Now the market is becoming more fragmented and suppliers (like United, American, and the major hotel chains) want you, as an OTA, to access the inventory via their site directly rather than via a clearing house.  However, as of April 2024, the GDS systems are still significant players in providing inventory, so using their infrastructure will suit our purposes.

As a start-up OTA, you'd likely want to use one of these aggregate solutions to provide you with a full-service offering rather than chasing down a bunch of suppliers to give you good deals before you have any leverage.  They give you instant inventory while you plot your next move to corner the travel market.

As a fake start-up OTA, I will be using whatever is minimal cost or free to fulfill our proof of concept purposes.  If this involves creating a fake data framework because the other options don't provide enough free or cheap options, then that's what we'll do.

And, again, if this isn't obvious, this is work that requires a developer or someone with sufficient tech savvy to build and maintain.

Ensure Mobile Compatibility

The comments for Develop your website by and large apply to this section.  On the front end (user-focused part of the site), the mobile components are a completely different set of services that serve similar functionality to your main website but are built for mobile platforms by developers with specialized skill sets.

There are debates within technology circles regarding the need to build specific mobile apps or rely on responsive web design (sites that automatically fit the browser you're using regardless of device) and progressive web apps (sites that change their behavior depending on your device and internet connectivity).

I don't plan on addressing that debate.  Your mobile strategy should be based on your customers' use patterns.  For the purposes of our project, I will rely on a mobile web strategy (i.e. no mobile applications) because I want to demonstrate what we can develop simply, even if it means making certain precarious trade-offs.

Until next time my human and robot friends.