--
Being a tech driven agency, as Lab Digital we are mainly focussed on delivering best-in-class digital solutions for our clients. The introduction of a diversity of AI services in the past years shows the tremendous possibilities that are now available. As a result, we are now using the capabilities of AI on a daily basis: for our employees, our clients as well as the services that we are using.
In a lot of areas within our jobs AI can be used to boost productivity. Our developers have a crucial role in developing solutions for our clients. ChatGPT can already be asked almost any question concerning writing, improving and debugging code. As well as ChatGPT, all our developers can use Github Copilot which is natively integrated in the development environments. It does real-time suggestions based on the project’s context and style conventions.
The advanced usage of the AI tools raises the recurring question: can it eventually replace developers? Will developer jobs become redundant? Or be replaced by AI engineers? We argue no, at least not at this moment. The solutions we develop for our clients are heavily context-specific, such as using unique business logic, integrating with existing environments or operating at scale. AI cannot easily take these elements into consideration, which makes its output in most cases not valuable enough.
However, of course, AI does help to improve our productivity. Do you want to write code where you already know it has been done thousands of times? Why not let it be written by AI? This way you have more energy available to focus on the cool, complex work, instead of writing “boring” code over and over again. In other words, we have more time to spend on adding value to the solutions we deliver for our clients. Here AI can really help, creating a win-win situation both for the happiness of our developers as well as added value for our clients.
We can even create more value for our clients by introducing AI capabilities in our solutions. In different projects we already have seen clients exploring the possibilities to improve their content with AI. We have seen two common ways where AI integrates within our solutions: for providing translations, and enriching content.
Translations
Recently, we developed a way to use the APIs behind ChatGPT (the OpenAI API) to translate content or product data. In the projects where we work on global e-commerce (MACH) platforms, often product data needs to be translated to different languages for all the markets where products are sold. While expanding to new markets, translating all the content by hand can be very time consuming and expensive. In the past our clients have relied on (expensive) translation services, with varying results. We see great translations for content now as a result of the wide-availability of AI technology.
2. Enriching content
Besides translations, we see that clients are looking for ways to generate or improve actual content. Especially when clients are making their first steps in the e-commerce domain, it can become too time consuming and expensive to generate SEO-friendly product descriptions for all their products. Normally, clients have a set of product data available that can be combined in a smart way to generate product descriptions with AI.
As in almost every new integration, we have experienced challenges with AI technology as well. For instance, we saw that when you have limited or low product data quality, AI doesn’t automatically fix that for you. We saw that the OpenAI API just returned that it for instance “can’t translate it” or “can’t generate something without nothing” so we had to build some exceptions to ensure that bad quality product data doesn’t end-up in the actual e-commerce platform.
Additionally, when working with a huge amount of variants with almost the same product data, you can end-up in a situation where the same translation or description is generated for multiple items. Of course, in that case you don’t want to send requests to OpenAI over-and-over again, saving time and money. To prevent this we built a smart way of hashing and saving the translated product data in our own shadow database. If we see that the hash of the product description is already available in the shadow database, we can just immediately use it.
We see that a couple of the services we use are introducing more advanced capabilities based on AI. These result in more efficient ways of working or enabling new features with minimal effort.
Especially in the search and merchandising area, artificial intelligence capabilities were already used for a longer period of time. Services like Klevu and Algolia are using artificial intelligence in such a way that they can provide advanced capabilities that go further than just the search bar with suggestions that every e-commerce platform has. For instance, Klevu has introduced MOI: a chat tool that works the same as ChatGPT for increased search and product discovery.
Besides this, we are using Sentry as our main tool for monitoring issues within applications. They recently introduced AI suggestion solution as an advanced feature to find the solution for the issue that’s appearing. Even though it’s currently experimental, it looks like a promising way to speed up detecting and fixing issues. This can further improve developer productivity and decrease the time to recover from incidents.
We expect that AI tools like ChatGPT will further develop in the coming years, with AI models improving and source data becoming more up to date and reliable. As an agency this will further improve the way we work with AI technologies and will even increase the value we add for our clients. It is certain that AI is here to stay. Without a doubt new advances will further improve the way we work with AI technologies and will increase the value we add for our clients as well as the efficiency and happiness of our developers.