One of the most frequently cited arguments against low-code platforms is the concern that such platforms cannot, or at least cannot fully, implement complex features such as artificial intelligence and machine learning applications, complex calculations and specific algorithms.
Such arguments are used to dismiss low-code platforms as ineffective and unsuitable for corporate use, and to exclude them from further consideration. This reflects a fundamental misunderstanding of the application of these platforms. They should be thought of as open integration platforms rather than a comprehensive solution.
In other words, low-code platforms provide many of the features of a traditional corporate application, and they do so quickly and efficiently. The high number of licensees allows the platform creator to maintain these features to a significantly higher standard than is possible with an individual development – a situation comparable to large cloud providers, who are able to invest substantially more resources in the security and maintenance of their features than most companies can do on their own.
At the same time, there are certain features that a low-code platform cannot or is not interested in providing. Let’s take the example of AI image recognition. Expert systems in this field have already reached a high degree of maturity, making it superfluous to redevelop the relevant features as a service on a low-code platform. This is not a criticism of low-code platforms: the creators in fact intend for such features to be integrated via API.
Practical example from the insurance sector
Here is an example: an application to categorise images from insurance claims can largely be implemented using a low-code solution. The registration and login processes, the claim form and the response to the end user are all standard features that are perfectly suited for implementation using a low-code platform. However, the insurer will come up against the platform’s limits if, for example, they want to process the images from the claims in the background and then make the payout either fully automatically or following assessment by an expert, depending on the level of damage. Evaluating an image using AI and determining the level of damage is not possible in a low-code platform. Integrating an expert system such as Azure Cognitive Services or AWS AI Services via an API call would be an appropriate solution here. During this process, the images and any relevant contextual information are sent to the expert system for processing and the result returned to the low-code application for further use – for example, to inform the end user whether their claim had been accepted or will be sent to an insurance expert for further evaluation.
The platform as an integration solution
Another potential application for cloud integration would be the use of Azure Functions or AWS Lambda to carry out complex calculations. This would allow users to take advantage of straightforward horizontal and vertical scaling and to implement computational logic in the programming language most suited to the specific application. This architecture could also include software developers able to engage with the material at a deeper level and to draw on their own tools and development process.
Many business processes require document management systems (DMS). Although it is possible to implement these using low-code, companies are often already using a specialist system. An API call allows these to be easily connected to a low-code platform to avoid implementing the same features again.
This ability to connect other services effectively removes the limitations of a low-code platform and enables users to enjoy its benefits without the disadvantages of limited functionality. In this architecture, the platform serves as an integration solution, uniting a range of features from systems that already exist or have been developed or purchased by a company in one central location. It enables users to create optimal software solutions for digital business processes through faster and more personalised application development.
Under the citizen developer model, the IT department not only handles governance and empowers its own employees, but also makes expert services available. These can be integrated and used by citizen developers as needed to build the precise logic that their use case requires.
This proposed architecture has been given additional impetus by the rapid developments in the public cloud field. In future, companies will obtain ever more services from cloud providers, whose market position allows them to drive innovation significantly faster than is possible with individual development. The low-code platform can then be used to integrate and orchestrate these services. Microsoft Power Platform, for example, is taking precisely this direction, primarily by making components available that help to integrate the whole Microsoft and Azure universe. The software development of the future is likely to continue in this vein, with low-code platforms gaining increasing importance as centralised integration hubs.
Silvan Stich has taken on the role of Head of Application Platforms as of January 2022. His focus is on team leadership and he is responsible for the development and advancement of the Application Platforms topic within the Cloud Practice. Silvan gained experience with application platforms as a project manager in various projects and bid phases. The great flexibility paired with fast implementation inspires him for the technologies.