The GenAI and low-code no-code revolution

The adoption of low-code no-code (LCNC) platforms has burgeoned in recent years, empowering businesses to develop applications sans extensive in-house coding nous.

It is a landscape evolving rapidly with the emergence of generative artificial intelligence (GenAI). LCNC leverages intuitive visual interfaces, pre-designed templates and drag-and-drop functionalities to materially accelerate the application development lifecycle. It is impactful for organisations streamlining processes, cutting costs and bridging the software skills gap. 

The uptake has been such that Gartner predicts up to 70% of new applications will originate from such tools and platforms by 2025. Forrester values this market at $13.2 billion, growing at a robust 21% clip annually since 2019. Adoption has been widespread, with 87% of enterprise developers now using such tools or platforms. 
 
GenAI will accelerate this trend, driving up market value to $50 billion over the next four years. 

While GenAI is reshaping the software landscape by enabling systems to autonomously evolve and adapt, LCNC platforms and Large Language Model (LLM) code generators/co-pilots are democratising software development and fostering digital transformation at an accelerated pace. The adoption of GenAI is also triggering big change in LCNC platforms. 

GenAI generates content, including code, when prompted. It is also being used for documentation, test-case generation, test automation, code optimisation and refactoring. For software developers, it offers new capabilities and tools to enhance workflows and offers a new way to develop quality software. 

With a conversational interface, GenAI allows developers to tap into a vast body of knowledge, easy access to code snippets and brainstorm in a more collaborative manner, allowing code development in days instead of weeks. Gen AI shines in quality assurance and verification processes. By assisting with code reviews and early issue-detection, it improves code quality and hastens testing cycles. Automation features further cuts the time for testing, while anomaly detection capabilities help identify hidden defects and potential issues early in the development process.

Risks and challenges 

While both systems have advantages, there are risks as well that require adherence to relevant laws and policies. 

Ethics, bias mitigation in algorithms, data privacy are paramount here. The surge in automation and interconnectedness wrought also elevate cyber risks, so guardrails are crucial.

Organisations also need to comply with regulations, such as the General Data Protection Regulation and the California Consumer Privacy Act to mitigate legal risks around data handling and AI utilisation and increasing regulatory requirements surrounding GenAI. 

Risk mitigation 

Companies must fully understand their ‘data protection obligations’ and approach data protection ‘by design and by default’. 

One way is to restrict direct access to application programming interface (API) and replace with a facade with internally hosted service to scrub personal data, flag misuse and abuse, and offer internal auditability.Here, users will call an internal API that validates the request and passes it through to an LLM. The facade API should be monitored and users authenticated through a single sign-on, or an authentication tool. 

Another option is to host own LLM or GenAI solution (open source) to preclude introduction of vulnerability and help models adapt to organisational data.Open-source, high-performance models such as Llama, Mistral and DBRX can be trained (quantised) at lower cost and provide greater transparency, control and flexibility. Small open-source models coming up can be trained on a single graphics processing unit (GPU). Open-source, locally hosted models will significantly reduce the risk of leaks, including of personal data.

Industry trends 

Industries are leveraging GenAI to experiment and innovate rapidly — with human oversight and expertise — for both open-source LLMs and proprietary models. 

Low cost, flexibility, and transparency make open source attractive for developers to finetune models for niche applications beyond coding to content development and knowledge management to boost efficiencies while curbing risks.Feedback from pilots indicates GenAI has been instrumental in streamlining code generation, debugging, and ensuring code consistency. GenAI can revolutionise software development and when integrated with LCNC tools, can drive tangible benefits.

Conclusion

The convergence of GenAI and LCNC heralds unprecedented opportunities for innovation and efficiency. They empower ordinary users to become creators of applications. As software development becomes more accessible, individuals from diverse backgrounds, regardless of technical expertise, can craft their own digital solutions and tools.  

Zak Murad

---

Zak Murad is Chief Technology and Information Officer at CRISIL Ltd.