It is time for another look at AI and low code as a call back to one of my previous blogs. The blog focused on how AI was helping with Low Code through the co-pilots and other capabilities. These capabilities are being introduced to us at a high pace. There is undoubtedly a case that can be made that low code also helps in AI adoption within the enterprise so this time around we will look at it from that perspective.
From a developer standpoint, we look for repeatable and reusable patterns, frameworks, and libraries to expedite our solutions, particularly in complex situations where the core solution alone might take days if not weeks to resolve and build up. Consider the recent trend of web applications that developers can swiftly utilize templates to get as close to meeting requirements as possible, enabling them to focus on new functionalities and a variety of customizations. When such capabilities are available, it’s highly likely that developers will adopt and remain loyal to the tools that simplify their work and lives. As a result, the prevalent use of various frameworks and systems today is a testament to developers and their organizations making well-informed choices and “adopting the way” behind specific technologies such as .NET or Java.
And so, we are in an era of diverse Generative AI flavours, each offering unique ways to leverage their capabilities (such as through API calls), while using distinct patterns to extract information. Take, for instance, the differences in the call structures to obtain data. Among the popular ones, like Open AI, there are subtle nuances that professional developers must consider, as effecting training is highly recommended but not necessarily straightforward or user-friendly. While the sample tools displayed on these apps do make it look easy but they strategically bait adoption. In essence, they follow what I would posit as the best way to solving problems: transforming the capability into a No Code/Low Code experience, where the “low code” element comes in the form of “Prompt Engineering” a.k.a. “how to talk to the Robot master to get the answer you really, really want.”
I would recommend that you look at your current low code tool and ask around (Chat GPT via Bing, perhaps) to discover what sample apps, patterns or pre-built connectors exist for communication with Open AI or other alternatives. You will be surprised to see that, in just a few months, the main vendors have made customizing the interactivity with Generative Chat systems much easier through No Code Low Code approaches. This progress has led to a point where the user’s journey towards “citizen development” potentially overlaps and then becomes a more trusted path towards enterprise adoption.
Solving problems is easier to measure than evaluating the time saved at work or productivity gains from daily interaction with those bots, especially when the assistance is generic (as in not using built-in functionalities for help or during the initial stages of using tools like M365 Copilot). The key lies in measuring these impacts because when organizational decision-makers observe progress aligning with their objectives, the likelihood of increased investment and support rises naturally, fostering can see the needles move in directions they like, the more likely investment growth from within.
Disclaimer: Yes, generative AI was used in this article but I can honestly say it solely to double-check on the two “references I made to “alternatives. This verification reaffirmed my existing beliefs and understanding. Hopefully, there is no need to cite any specific resource other than suggesting you go ahead and ask the same questions and to validate your case, as it may be one of those alternatives. The search should propel you further along the journey, which is a good thing, regardless of the path you have chosen.
Also, while you are looking, make sure to ask for those samples and patterns as a standard procedure before embark on building something, just like the great idea for the wheel!