Low code is far from new and has struggled to gain widespread enterprise popularity. Yet the arrival and adoption of artificial intelligence (AI) is not a threat to low code. The principles of low code are providing a safe harbour for enterprise AI development.
“AI unleashed will destroy the software industry. Goodbye to SAP and Oracle and all the software industry. That is absurd,” says Diego Lo Giudice, principal analyst at Forrester. “It would require AI perfection.”
Speaking at the OutSystems One Conference, the technology analyst poured cold water on claims that AI will do to the enterprise software industry what the internet did to paper rounds and sending letters.
This is not to say that AI will not impact enterprise software development, but as with much surrounding AI, the case for its abilities is overblown and not rooted in the real business needs of organisations. Giudice believes AI will change the development cycle of enterprise software as it reshapes planning, requirements gathering, epics (the large, high-level pieces of work that are too large for a single sprint or iteration), code generation, documentation, testing and code translation.
His research into the existing and potential impact of AI on software development has led Giudice to believe that software development lifecycle processes will be reduced: “With AI, we can create test cases from the requirements, so generative AI will change the software development lifecycle, and we cannot keep doing things in the same way.”
Giudice says the initial adoption of AI in enterprise software development automated each existing stage, but as developers become more comfortable with AI and begin to use low-code tools as part of their AI development, then they can use AI more powerfully. He cites an ability to cut across different processes; for example, code generation will create a call for testing of that code, and the testing will automatically create a call for deployment.
CEO of OutSystems Woodson Martin agrees with the analysis of Forrester and says this has shaped the theme of the low-code application development provider’s annual conference: “Customers and developers here have both the skills, background and existing assets in terms of data and APIs, and now the tools they need to go and lead the agentic AI agenda for their companies.”
CIO for OutSystems Tiago Azevedo adds: “Agents create great impact in processes and workflows, and I say that as a CIO who is seasoned enough to have been through all of the mega trends.”
Low code’s role
As AI becomes part of the software development lifecycle, it is easy to imagine that low code will lose its place. “If I can use generative AI, why do I bother with low code? I can just vibe code,” says Bruno Martins, IT advisory partner with KPMG.
But industry watchers IDC find no signs of the developer community abandoning low code. The study Modernizing SAP for the future: low-code and AI enable enterprise innovation by Arnal Dayaratna finds that 40% of professional developers are using low code on a widespread or intermittent basis. Dayaratna writes that low code is benefiting businesses through increased collaboration that “bridges the gap between IT and business stakeholders, allowing for faster consensus and more targeted solutions”.
The IDC study focused on enterprise modernisation of SAP, the core system of record in many organisations. Technology decision-makers are worried about SAP modernisation when there are competing priorities for digital transformation, AI adoption and budget pressures.
However, if the SAP estate is not modernised it is open to cyber security risks and will hamper those AI adoption and digital transformation demands. For example, one charitable organisation has moved its legacy SAP estate over to a managed service provider (MSP) while it looks at a wider business and technology transformation.
Faced with these complex technological and business issues, IDC finds that organisations are turning to low code as a “strategic enabler for broad-based AI adoption within the enterprise. IT leaders in healthcare, financial services and education tell us they are migrating to a new SAP environment and will therefore have to write new code. This trend is underscored by recent IDC survey data showing that 30% of production-grade applications with AI or machine learning functionality are now developed using low-code platforms.”
The research found developers using low code to develop AI-based predictive analytics, computer vision and model optimisation: “This approach streamlines the integration of AI into core business processes and transforms AI-driven capabilities from a specialised resource into a standard across the organisation.”
Martins describes the role of low code in enterprise AI software development as the need for validation. AI may bring speed, but the modern enterprise needs more than just pace, with cyber security threats at an all-time high and customer sentiment a concern; robust applications are essential. As with the recent history of low code, organisations using the technology for AI development are reducing backlogs as they are utilising templates, KPMG finds.
With every software-as-a-service (SaaS) supplier talking up the potential of agentic AI, organisations face the potential for a complex architecture where it will be hard to understand dependencies between business processes and technologies. Again, low code has the potential to step in and help.
Legacy issues
In under two years, support for SAP ECC comes to an end. There is the option of continued support, but many believe this is an opportunity to modernise the organisation. For those opting for migration, the IDC study finds that AI can increase developer productivity by 40% through AI-generated code, testing and debugging.
“This productivity gain allows developers to redirect their efforts toward higher-value activities such as application design and architectural decision-making,” says Dayaratna.
A further benefit for organisations grappling with legacy systems is the way low code can help teams to collaborate, so business analysts, line managers and the IT team can coalesce around a business process and work together to modernise the technology, architecture and business needs.
Giudice at Forrester believes this trend will drive yet further productivity gains from development. As the models behind AI improve, then so will the quality of the AI code outcomes.
“It will change methods, handovers and discussions,” he says. That, in turn, will lead to changes in the enterprise application development platforms as they adopt a higher level of AI and prove their ability to deliver “secure, consistent, compliant, integrated and interoperable” technology.
It is not only the established low-code firms that are bringing the template methodology to AI enterprise software development. In our research, we found 15 new AI enterprise software development providers offering outcomes to the enterprise. These include AI agent development (Lindy), internet of things and API workflows (Node-RED), visualised workflow construction (Make), web app integration without code (Zapier), sales and support automation (GPTBots.ai), large language model and workflow integration (Gumloop) and iPaaS platforms (Workato).
Organisations and their developers are delivering business outcomes. Berlin-based n8n was used by UK mobile telecoms operator Vodafone to develop a security orchestration, automation and response (SOAR) outcome. Since August 2024, Vodafone has created 33 workflows using the low-code AI technology, which they believe has saved 5,000 working days and avoided £2.2m in costs.
Austin, Texas-based online health and medical retailer Truemed had a need for internal application development but was finding it too complex to guarantee a good return on investment (ROI). It adopted the Lindy platform to ensure it could cost-effectively build applications but not have to use too much of its internal software engineering resources or hire additional staff. Truemed describes AI low code as a junior developer.
Smart Charge America, the EV charging services provider, has used the Zapier low-code AI technology to automate the quoting process, which they say has increased the speed of installation of EV charging points by 75%, saving the firm 145 business days a month.
Also in the rechargeable sector, GP Batteries of Hong Kong adopted GPTBots.ai to develop customer service automations that have taken on 50% of customer responses and reduced the support service operating cost by 50%. This has given the business a 24-hour, seven-day-a-week customer service team.
Organisations develop their own enterprise applications when the vanilla out-of-the-box solution cannot satisfy the needs of customers or business lines. Low code brought governance and rigour to application development, and its template nature ensured the business always owned and had access to the code base to maintain the application long after the developer’s career.
As organisations adopt agentic AI, like low code, to satisfy the needs of customers and business lines, the early adoption of low-code principles could ensure organisations remain in control of their agentic future.