Contributed ArticleUAE businesses, and many of their GCC peers, now operate in an ecosystem of pressures. Customers want the best digital experiences right now, and more tomorrow. Regulators insist on the highest standards of privacy and security. And governments want innovation that will shine on the world stage for all to see. The future, it seems, is made for those that can deliver trusted, quality digital experiences at breakneck speed.
Regional technology leaders faced with this challenge must deliver against a backdrop of skills gaps. More and more software from a talent pool that does not have the knowledge to deliver it seems like an unrealistic demand. Even if the skills are there to meet some of the demand, the CIO may not have enough coders to meet all of it.
When non-technical staff can code, it opens several avenues of operational enhancement
As you read this, the UAE’s National Programme for Coders is in full swing, aiming to attract 100,000 software professionals and plug the country’s IT skills gaps. But the initiative, bold and visionary as it is, will take time to achieve its goals. Meanwhile, businesses have customers to appease, and the burden will fall on the IT department to produce the digital goods. Under normal circumstances, IT would farm at least some of the work out to a contractor, but today that is unnecessary.
Tools now exist to allow non-coders to build solutions without writing much, if any, code. A combination of cumulative industry knowledge and the advent of cloud computing has given us the low-code development platform, LDP, an intuitive work environment made for business-oriented users where point-and-click and drag-and-drop replace hours of syntax-aligned typing and debugging.
Non-coders learn new skills and get to perform more challenging tasks
Low-code and even no-code tools have been around for a while and are already proving to be effective accelerants for digital transformation. When non-technical staff can code, it opens several avenues of operational enhancement. Non-coders learn new skills and get to perform more challenging tasks. Professional coders, to whom these tasks were humdrum, get to concentrate on more complex problems.
The projects that are undertaken by non-coders have a quicker time to production and a greater success rate because the developer is also the end user, which means requirements do not get lost in translation during laborious requirements-gathering exercises.
Additionally, as more and more people across the organisation get to access and innovate with data, ideation and innovation increase. Low code even works with advanced technologies such as AI. Technology leaders who become aware of the benefits of low code can discard the traditional trade-off between off-the-shelf AI products and pure-build solutions, where they had to choose between empowering business analysts at the risk of losing competitive edge and empowering data scientists at the risk of poor production outcomes.
But today we know a little more about the upsides and downsides of such choices. We have come to realise that a culture of collaboration, involving people from different parts of the enterprise, is necessary for the success of an AI project. Business analysts, data scientists, IT administrators, and business users all have a role to play in introducing Everyday AI.
For low code to work, especially in an Everyday AI setting, the technology space must be easy to understand for the non-technology professional. IT should engage in a spring-cleaning venture that eliminates data silos and establishes a unified analytics-ready environment. Next, stakeholders need to have a meeting of minds to uncover the most appropriate internal use cases.
No- and low-code platforms are not just for citizen developers. Professional coders can use them to speed up development of solutions too complex to assign to a non-coder. Each use case can be assigned a developer whose skills are appropriate to the challenge.
Sometimes data science projects will be resource-heavy and citizen developers can work on parts of the solution where it makes sense. This is an immensely effective use of resources and with the right resource strategy, can accelerate development of even the most complex projects. Such tasks include smart data ingestion, data cleaning, and dataset merging.
In some instances, the citizen developer may even have the opportunity to create new machine learning models, apply model assertions to capture and test known use cases, and conduct what-if analyses to interactively test model sensitivity. All of these activities can be significant boosts to job satisfaction for business users and subsequently to talent-retention rates.
Meanwhile data scientists, software developers and other IT roles will have room to innovate, applying their skillsets in more value-adding areas. Low-code tools not only help citizen developers to alleviate the burden on technical staff. Technical staff can use them directly in a number of ways.
For example, LDPs can help data scientists with model maintenance. Visual, collaborative interfaces for data pipelining and preparation, as well as model training and support for MLOps, allow data scientists to scale their models more easily. LDPs can deliver transparency and traceability and reduce the likelihood of failure or disruptions to operations. When it comes to Everyday AI, LDPs deliver the visibility needed to gather clear performance metrics for models and draw a clear line between design and production environments.
Another boon to the data scientist is code reuse. Low- and no-code tools all but eliminate the repetition that would normally plague their daily schedule, especially where code silos obfuscate workflows and tasks are unnecessarily duplicated. LDPs can be repositories and catalogues for code that allow data scientists to see at a glance what tasks have been completed and which are pending.
On the quick march to digital transformation and Everyday AI, low-code and no-code development are indispensable elements of a larger strategy — one in which each skillset, technical and non, is utilised to the fullest.
