Digital Adoption Platforms (DAPs) have become essential tools for helping employees navigate complex enterprise software such as CRM, ERP, or HRMS systems. Companies like LemonLearning have shown how digital guidance can reduce training costs and accelerate onboarding. However, building and maintaining these interactive guides still requires extensive manual effort. Leveraging Large Language Models as virtual assistants is an appealing alternative, yet without a structured understanding of the target software, LLMs often hallucinate and produce unreliable answers. Moreover, most production-grade LLMs are black-box APIs, making fine-tuning impractical due to the lack of access to model weights. In this work, we introduce a Graph-based Retrieval-Augmented Generation framework that automatically converts enterprise web applications into state-action knowledge graphs, enabling LLMs to generate grounded and context-aware assistance. The framework was co-developed with the AI enterprise RAKAM, in collaboration with Lemon Learning. We detail the engineering pipeline that extracts and structures software interfaces, the design of the graph-based retrieval process, and the integration of our approach into production DAP workflows. Finally, we discuss scalability, robustness, and deployment lessons learned from industrial use cases.

Over the last several months, OpenAI, and ChatGPT in particular, has shown what's possible with a user interface built on top of a large language model that can answer questions and create code or pictures. While that alone is remarkable, we can also interact with and adjust the byproduct by having a conversation of sorts with the AI. It's amazing really, but think about how transformative this could be by applying it to the enterprise applications you use on a daily basis. What if you could build an interface on top of your existing applications, so that instead of pointing and clicking, you could simply ask the computer to do a task for you and it would do it, based on the applications' underlying model or your company's internal language model. That would be a huge leap forward in computing.

There's a lot of hype around machine learning, but what does it really mean in the context of enterprise software? How does it work, where is it adding business value today, and what should we expect from it in the future? Let's start with some definitions. Artificial intelligence (AI) is an umbrella term that includes machine learning (ML), deep learning and cognitive learning. The part most relevant to enterprise software is ML, which in this context is the ability to create automation through AI algorithms.

There's a lot of hype around machine learning, but what does it really mean in the context of enterprise software? How does it work, where is it adding business value today, and what should we expect from it in the future? Let's start with some definitions. Artificial intelligence (AI) is an umbrella term that includes machine learning (ML), deep learning and cognitive learning. The part most relevant to enterprise software is ML, which in this context is the ability to create automation through AI algorithms.

Think for a moment about how organizations are evolving. They all want to be lean, fast and specialized. They're trying to become more adaptable, intelligent and resilient. Wherever you look, it's a pretty consistent pattern. Now, think about the main trends in enterprise software.

For many technologists, developing software for the enterprise holds a certain appeal. For example, you get to build things used by millions (potentially even billions) of people. Second, it's an area of software development that can prove quite lucrative, provided you build something used by companies all over the world (and you're compensated proportionally for your impact). Third, enterprise software often attempts to tackle complex problems--an intriguing prospect for technologists who really like puzzles and intricate solutions. While builders of consumer software and video games might think that enterprise software is boring, it's actually quite exciting to those of a certain mindset.

Data is growing by leaps and bounds, the convergence of extremely large data sets both structured and unstructured define Big Data. The increasing awareness of the Internet of Things (IoT) devices among organizations and volume, variety, velocity and veracity at which data is generated have caught the attention of the enterprise in a bid to enhance digital technologies and guide digital transformation. Analytics Insights eliminates that the big data market size will grow at a CAGR of 10.9%, globally from US$ 193.5 billion in 2020 to US$ 301.5 billion by 2023. This region is witnessing significant developments in the big data market gaining remarkable traction in the BFSI industry vertical. Numerai is the world's first hedge fund, to predict the stock market.

SOMEBODY LESS driven than Tom Siebel would have long since thrown in the towel. In 2006 the entrepreneur, then 53 years old, sold his first firm, Siebel Systems, which made computer programs to track customer relations, to Oracle, a giant of business software. That left him a billionaire--but a restless one. In 2009, a few months after Mr Siebel had launched a new startup, he was trampled by an elephant while on safari in Tanzania. When, a dozen surgeries later, he could work again, the enterprise almost went bankrupt.

Conversational user experiences, in the form of chatbots and voice interfaces, are overtaking many of the traditional ways in which we interact with machines. Since the rise of computers, human-machine interfaces typically had some form of Graphical User Interface (GUI) which enabled direct (if limited) interaction with devices and their programs, for instance via software installs, mobile apps, and web-based applications such as Software as a Service (SaaS). No matter how "beautiful" the respective interface, this GUI is now more and more replaced by a Conversational User Interface (CUI). Other still evolving interface styles are less text- and voice-driven, and therefore limit the messaging element to certain basic functions such as taking photos with the blink of an eye (smart glasses or smart cameras such as Blincam can do that today) but will eventually allow for richer interaction gestures (see project Soli). When coupled with an input-output feedback loop, so-called bionic lenses also hold a promising future.

That Artificial Intelligence will replace the need for a human workforce, is a misconception. Rather, AI is an effective tool that can support a business to build a more efficient workforce and assist with cost-saving initiatives. This is the view of Wessel Oosthuizen, Senior Manager, TMT, Deloitte Africa, who says that skills and capacity building in the field of AI, as well as critical thinking and a workforce that understands how to leverage this technology – are key for businesses to remain relevant in a rapidly changing world. "Individuals with finance, engineering or computer science backgrounds – who understand both the possibilities that AI presents and its limitations – are priceless. They combine their subject matter expertise, critical thinking, data analysis and innovation, with AI tools and techniques to help their organisations make the most of AI and similar technologies" said Oosthuizen. His sentiments are echoed in Deloitte's recently released 18th edition of its Technology, Media & Telecommunications Predictions (TMT Predictions), which finds that while enterprise software and cloud-based development platforms can provide effective gateways to AI, they are not a substitute for having at least some technical AI talent, in-house.