Generative AI (GenAI) is here, and with it come exciting new capabilities in the world of ecommerce. With it also comes the familiar question of “should I build or should I buy?”
One example of what we’ve seen are AI Shopping Assistants, allowing shoppers to communicate complicated requests to an ecommerce site like “I’m going hiking in the Rocky Mountains next week, what weather-appropriate gear should I bring?” or “Give me some ideas for healthy recipes to cook for a family of four.”
Another is Attribute Enrichment, which automatically generates almost any new attributes desired for products, rapidly speeding up the previously manual human tagging required to correctly add attributes to products.
And these are just the beginning. Just about everyone in the industry knows we’re still only scratching the surface of what’s possible.
This all brings us to a familiar question from the early days of the internet that rears its head around new technology every few years: what should you build and what should you buy?
Navigating the Build vs Buy Dilemma
The question comes up every few years because emotions tend to run high around it.
On the one hand, the world has a lot of bad software and a lot of good engineers. It’s easy to find examples where someone at an organization purchased bad software, and then good engineers are asked to integrate it. In the end, they spend more time integrating it than they would have spent just building a better version themselves that’s more tailored to their use case.
On the other hand, I remember working as a search engineer a decade ago. I was tasked with making improvements to internal search algorithms at ecommerce sites. All the while I knew that the biggest ecommerce companies had one hundred times as many engineers to devote to the problem and could, therefore, develop better, more intricate solutions faster. It was an economies-of-scale problem.
I’d be proud of something I wrote because it would win an A/B test, but then realize that it felt like I had built a kayak while those larger companies had already created a battleship. The kayak worked perfectly well, but it’s not what you want when challenging a battleship.
My circumstances weren’t unique. There were thousands of other engineers like me, working on the exact same problems and reinventing the wheel at other ecommerce companies, while our largest rivals could use the sheer size of their teams to pull further and further ahead.
It’s been years in the making
This problem existed originally in server warehousing, too, before AWS came along. Companies were building their own server warehouses, each one hiring teams to work on almost the exact same problems around managing a fleet of servers.
These days, aside from a few holdouts, this is thankfully mostly a thing of the past. It took a lot of blood, sweat, and tears, but you’re now much more likely to find an ecommerce company using a cloud like AWS or Azure for its server warehousing and a cloud-based SaaS like Constructor to power their product discovery and search than you are to find these technologies being built from scratch.
And you’re much more likely to find the engineers who would have worked on building software like this in the past now genuinely innovating on new technology rather than reinventing wheels.
As a slight aside, this is exactly why at Constructor, we use SaaS and open source technology wherever it does a good enough job for our needs. We want to spend our time focused on building differentiating technology for ecommerce that doesn’t currently exist. We much prefer to purchase the building blocks from someone who is an expert in that field.
I’m obviously biased on this point, but I challenge anyone to call this anything but good.
Innovation as the Key to Success in the Age of GenAI
And now we have the same questions starting to arise around GenAI: Should you build your own LLMs, and should you build the technology to power them yourself? Similar to server warehousing and search, it would be a fun science project, but the industry is changing too rapidly to confidently go it alone.
GenAI is a fast-moving field, and there's a lot of potential value for the fastest movers who know how to capitalize on and leverage the currently available technology.
While some ecommerce companies reinvent the wheel and try to build their own versions, others will innovate on top of what already exists — getting to user value first.
We’ve seen this movie before. We know how it ends. Focus on innovation. Don’t reinvent the wheel.