Credit: Fatos Bytyqi
Creating the digital infrastructure that runs the planet is hard work. There’s a reason why coders are paid so well—and it’s the same reason that everyone is so entranced by solutions that are said to be “low code” or even “no code.” The cumulative savings in time and money could be astronomical.
The good news is that the trend is real. Programmers have been building new tools like compilers and debuggers that simplify their work for decades. Computer scientists have started to develop tools that will save us all many hours of endless tedium. The latest batch of low- or no-code tools may be the best ever.
The newest systems let us access the best of artificial intelligence in a superior user interface. Developers don’t need to type endlessly in text editors anymore; they can string together cute icons and leave the details to the machinery behind the curtain.
Low code and no code tools are also bringing artificial intelligence into data pipelines around the enterprise. Companies that can't afford to hire their own AI boffins can just fire up some low code tools that have the AI integrated under the hood.
The not-so-good news is that the dream of solving all our IT chores with a single click is unlikely. The power of these tools to save time is very real, but they often require guidance from smart humans. None of them will magically transform someone who failed computer science into an IT hero. What they can do is make the work faster and easier for those of us who grok programming. It really does help to have a feeling for what’s going on under the hood.
Let's take a look at the top seven technology areas where low code and no code solutions are changing the face of IT.
Robotic process automation
The name doesn’t really fit because there are no anthropomorphic machines waving their arms around saying, “Danger Will Robinson!” Yet it has somehow stuck. There are easily several dozen platforms designed to simplify the paper-born tasks that haunt accounting and compliance departments. Banks use these tools for opening accounts. Warehouses use them for bills of lading.
Machine vision tools for optical character recognition are essential to data pipelines. These AI routines can scrutinise government-issued identification or locate the invoice number on a bill.
Many companies also adopt robotic process automation (RPA) as a way to modernise their old infrastructure. The same people who can tell you where to send every form for approval can use the low-code editors to create pipelines that move digital forms between offices, factories, and warehouses. The low-code layers can interact seamlessly with many of the protocols and file formats used by legacy systems. They do much of the work of translation and reformatting with minimal configuration.
Some of the top RPA systems include UIPath, SAP, Appian, and Automation Anywhere. The area is also rapidly expanding as some companies merge and others integrate RPA features into existing automation platforms. Microsoft’s Power Platform, for example, integrates RPA with similar chores like creating business intelligence reports.
Business process automation
Another common term used to describe low-code systems is business process automation, or BPA. Another term is “business process management,” or BPM. There’s not much difference between the applications that live under this heading and systems defined as robotic process automation. Functionally, RPA, BPA, and BPM all link together various tools with very little coding. There’s plenty of overlap between them.
Some of the best-known tools include Zapier, Creatio, KissFlow, and SnapLogic. Some of these tools offer a particular focus; AirSlate, for instance, wants to help with document flow through the enterprise. It offers tools for automating document creation, version control, and authorisation.
Low code and no code AI tools
While many companies are integrating artificial intelligence algorithms into their own low-code products, some are making a business of selling AI tools. The theory is that developers will use the AI as an assistant who eats batches of code for lunch. A well-known example, GitHub launched GitHub Copilot after training it with OpenAI’s tools on the bazillions of lines of open source code stored in its own servers. Others are turning directly to OpenAI, whose APIs offer direct access to models like code-davinci-002 and code-cushman-001, which are said to be quite capable with many of the most popular languages including C#, JavaScript, Go, Perl, PHP, Ruby, Swift, TypeScript, SQL, and Shell scripting. Some open source versions include GPT Code Clippy and SecondMate.
While the most prominent tools endeavor to write large blocks of code that tackle an entire programming task, others are less ambitious. Captain Stack, for instance, discreetly uses a search engine to locate good answers on forums like Stack Overflow, then modify them into code suggestions for your editor. Similar projects are Clara Copilot, YouCompleteMe, and Kite.
Opinions about the quality of AI code contributions vary. Most are amazed that machines are able to stitch together solid solutions that come close to delivering exactly what’s needed. Still, machine mistakes require human attention.
Super applications
Some developers are automating their workflows by taking advantage of the opportunities for integration that show up in what were once considered basic applications. AirTable, for instance, is a combination of a database and a powerful spreadsheet-like interface that makes it relatively easy for spreadsheet users to start creating more sophisticated software.
Many of the major platforms are slowly integrating their tools to do the same thing. Companies like SAP, Amazon, Microsoft, Salesforce, and Google are adding so many links between their products that it is fairly simple to create elaborate workflows with minimal code. As one example, Amazon Web Service users can write Lambda functions that knit together many of their products. Google back-end tools are often tightly integrated with their office products, resulting in spreadsheets that can respond to other software or initiate events.
Web applications
Many of the most well-known low- and no-code solutions are used by people who are building web applications. Tools for editing web pages and placing pictures often include features for tweaking the CSS that defines the look. They also often allow adding basic JavaScript functions for common tasks like selling merchandise or collecting forms.
Companies like SquareSpace, Wix, Strikingly, Webflow, SITE123, WebNode, Web.com, and Weebly are just a few of the major options for creating basic websites with a bit of automation. There are also good open source solutions like Drupal, WordPress, and Ghost, which are also supported by companies that offer hosting services and customisation. Some focus on specific niches, like Pixpa, which builds portfolio sites for artists.
Companies like Shopify, BigCommerce, Opencart, Adobe’s Magento, and Drupal Commerce are site builders that specialise in online stores. Many users are able to create elaborate stores with deep databases with no code. Those that require additional functionality can often write just a few lines and incorporate them into the web flow.
Some are also using these tools to build desktop and mobile applications. Frameworks like Electron, NW.js, Flutter, Neutralino, and ReactNative are toolsets that turn HTML, CSS, and JavaScript into something much closer to a native app. They often bundle an open source browser like Chromium with your code to build something that runs and functions like any desktop or mobile application. Some find that the results can be unwieldy and a bit slower than hand-coded C++, but they remain popular because all you need to use one is familiarity with building web applications.
Data storage
Many parts of data management involve triggering events, passing messages, synchronising data streams, and creating dashboards or reports. All of these tasks can be handled by tools. The tools themselves are becoming sophisticated enough to require little or no coding experience.
Databases were once complicated to install and tune for performance, but now companies like Oracle can easily slap the word autonomous on them. Companies like Amazon, IBM, MongoDB, Google, PlanetScale, and ExoDB offer hosted services that automatically manage installation and tuning.
Some companies are building hosted data lakes and data warehouses that can be integrated with data sources across a stack. They come with predefined functions that ingest the data, answer queries, and generate reports, all with very little coding. Some major options include Snowflake, Databricks, Cloudera, Panopy, and Dremio.
The work of building this infrastructure is creating large systems that often are classified with more general terms like business intelligence platforms. Microsoft’s Power platform, SAS's business intelligence infrastructure, and the products Tibco and Tableau are just a few examples.
General infrastructure
Many of the tools mentioned so far were built for a particular use, but not every job falls into such a niche. For the rest of the workload, there are good low-code choices for tackling general chores. Some of these tools are built directly for developers and adept users, and these generally combine a visual programming editor with a collection of back-end routines for accessing databases and remote servers with standard formats like JSON or XML.
Some of the most popular versions include SAPs Build Apps (formerly AppGyver, Make (formerly Integromat), Node-RED (formerly from IBM), Clutch.io, Mendix, Quixy, and Google’s AppSheet. Lansa, meanwhile, is a more niche solution that includes features for simplifying the work of modernising older IBM code.
Visual editors rely on a combination of dragging and dropping with some clicking to fill out predefined forms. The products are pitched as “low code” because there’s rarely much need to edit text files and worry about the idiosyncrasies of parsers with their demands for proper punctuation. Still, much of the higher level thinking is familiar. As a creator, you must think about the structure of the data and how it moves, even if the platform will handle many of the basic chores.
These products also often include pre-developed modules for connecting to common APIs using versatile protocols. If one of the more specific tools I've discussed doesn't fit the job, more general solutions are often the best choice.
