The sudden explosion of generative artificial intelligence has upended the low-code/no-code landscape, as it has done for so many corners of the software industry. I trekked to the NoCode Summit in Paris this week to get the latest story.

I found a surprising mix of vendors: some focusing on gen AI, but more often than not, the emphasis was more about where in the stack each vendor plays.

On display were front-end, back-end, workflow, and full-stack vendors as well as the AI-focused offerings. Here’s the lowdown:

Pinning down the terminology

My first surprise upon arriving at the NoCode summit was just how many “no-code” exhibitors were actually low-code vendors. I define low-code as a visual development platform that allows for hand-coding when necessary, thus supporting both business users and engineers. No-code, on the other hand, is exclusively for business users – no coding allowed.

Many of the vendors at the conference, however, defined these terms differently. For them, “no-code” means that coding is strictly optional (but still possible), while “low-code” requires some amount of hand-coding. For the purposes of this article, therefore, I’ll use the terms interchangeably, or punt entirely by referring to them as low-code/no-code tools.

There was also plenty of confusion about AI. Even though gen AI is the new kid on the block, at this conference, “AI” meant “gen AI.” Other forms of AI, including machine learning, computer vision and even natural language translation never came up in conversation.

I was also surprised at how many of the French companies played fast and loose with their English language support. They all sported buttons that switched the language, but the French screens all showed plenty of English and vice-versa. Sacre bleu!

Unstacking the stack

Because of their visual nature, low-code/no-code tools have traditionally focused on building apps and web sites – or in the parlance at the conference, front ends.

Front end

The front end-focused vendors I spoke with at the show included SAS GoodBarber, Pickaform SAS, Stacker Software Ltd. and Baserow B.V.

GoodBarber focuses on building mobile apps for IoS and Android, as well as progressive web apps or PWAs that run on all devices and browsers. They feature an integrated no-code approach that offers templates for customization rather than components that require assembly.

As its name would suggest, Pickaform focuses on web forms, as well as the processes that support them. The tool provides a spreadsheet-like interface and a simple workflow builder. Pickaform also offers fine-grained entitlements that enable users to secure individual field elements.

Stacker combines application templates and fine-grained permissions to support the no-code construction of specific types of applications. By restricting the apps to a fixed list — CRM, portals and a few others — Stacker makes good on the no-code promise that business users can build and configure apps with no technical help, although coders can still add bits if necessary.

Baserow tackles the problem of spreadsheet sprawl with a low-code app builder that exposes tabular data in portals, web sites and other app user interfaces. The company also allows app builders to use gen AI to fetch data to populate applications.

Back end

There is more to an application than its UI, however, which explains why low-code tools for back end development are becoming an increasingly important part of the low-code/no-code story.

Two vendors at the NoCode summit stood out for their support of back end development: BuildShip (Rowy Inc.) and Xano Inc.

For BuildShip, the back end includes APIs, automation and integration. Its product is a visual back-end builder that enables users to assemble blocks of JavaScript or TypeScript code. While BuildShip offers “back-end as-a-service” running on Kubernetes, the company thinks of itself as more of a low-code development tool.

Xano also provides low-code tools for building enterprise back ends. The company offers a no-code API builder that supports REST as well as its GraphQL-like Addons. Xano back ends include a Postgres database and run on Google Cloud by default, although customers can choose other options.

Workflow automation

Low-code/no-code’s visual boxes-and-lines metaphor lends itself to assembling workflows, so some kind of workflow creation and automation capability is built into many of the products at the conference.

However, two vendors stood out for their workflow support: Compose Software AS and Make by Celonis Inc.

One of

One of Compose’s key differentiators is the fact that it offers both a low-code tool for engineers, as well as a separate no-code tool for business users, each running on the same underlying technology. Both Compose platforms have built-in confidentiality support, making it particularly useful for building workflows compliant with GDPR and other confidentiality regulations.

Make offers a visual low-code workflow creation and automation platform that competes with integration-centric platforms such as Zapier and Workato. The company also takes on the integration-platform-as-a-service space generally with its more intuitive, low-code approach to building integration-centric workflows.

Full stack

Modern applications require both a front end and a back end to work properly, of course – but leveraging low-code to support the creation of both parts of the stack at once requires some heavy lifting.

Three vendors at the NoCode Summit, however, are choosing to make that lift: AppFarm AS, Convertigo SA and Caciopee sarl.

Appfarm’s low-code platform includes support for UIs, business logic, data and integration. App creators assemble JavaScript-based blocks into workflows, thus allowing coding if necessary. Appfarm also offers a “pixel-perfect” UI builder for building mobile apps or PWAs.

Convertigo also offers pixel-perfect UIs on top of JavaScript-based building blocks – and even supports AI-generated JavaScript for some functions. Convertigo stands out because its technology is open-source, thus mitigating concerns about vendor lock-in. It is also lower cost than Mendix and OutSystems, which it sees as its primary competitors.

Caciopee also competes with Mendix and OutSystems with Works, its mature full-stack offering. Though many of its customers have built complex applications on Caciopee, it remains largely unfamiliar to European and North American audiences. The company is based in Morocco and serves a largely African and Middle Eastern clientele.

Gen AI to the rescue

The holy grail of gen AI-supported low-code is being able to describe the application you want in your natural language of choice. Push a button, and out pops a fully functional, usable application to spec.

Unfortunately, this vision of seamless application generation is still just out of reach. Nevertheless, several vendors at the conference fell somewhere on their quest for this grail, including Latenode (NOCODE Ltd.), Adminix Solutions Inc., THINKEO, SAS and Mindflow SAS.

Latenode offers a gen AI copilot for developer support and documentation as well as AI-generated code for automating individual tasks. App creators then assemble these blocks using Latenode’s visual workflow tool.

Adminix provides a full-stack low-code application builder that uses gen AI to create initial drafts of workflows. Human application creators then use Adminix’s no-code tool to adjust those workflows to meet requirements before publication.

Thinkeo offers a no-code application builder for assembling gen AI-based agents into workflow-based applications (see my article on AI agents for some background).

With Thinkeo, application builders can run AI agents in parallel, say to compare results from different language models. It follows a graph metaphor that supports business logic among agents, including loops and other programmatic constructs. Application builders can also use retrieval-augmented generation or RAG to include various unstructured data sources as inputs.

Mindflow offers its AI-supported no-code application builder for security, cloud, and information technology operations professionals. Mindflow-based applications include AI agents that serve as collaborators with ops and security personnel, providing reasoning capacity to humans who always remain in the problem-solving loop.

It integrates with dozens of security and operations tools on the market, offering complete coverage of each tool’s APIs to help security and ops personnel break down the silos between tools to resolve issues more quickly.

What’s next for low-code and no-code

There’s no question that the NoCode Summit revealed a marketplace in the midst of a disruptive transition. The writing is on the wall: It’s just a matter of time until AI can handle all the tasks of building applications – even those tasks that visual development tools handle today.

Gen AI’s code generation capabilities are also disrupting the professional development or “pro-code” marketplace as well. I predict, therefore, that no-code, low-code and pro-code will soon collapse into a single AI-generated application space where the human effort will shift away from application construction to focus on how best to describe the applications that companies want.

If you look closely at this trend, however, you’ll see that there is a gen AI shell game going on. The skills necessary to describe an application properly in sufficient detail amounts to many of the same skills people need today to build applications, regardless of which tool they use.

Jason Bloomberg is founder and managing director of Intellyx, which advises business leaders and technology vendors on their digital transformation strategies. He wrote this article for SiliconANGLE. Mendix, OutSystems and Workato are former Intellyx customers.

Photo: Jason Bloomberg