Though still in its early stages, we are beginning to see some of the benefits and impact that AI can bring to the enterprise. In everyday work, tasks such as summarizing meeting notes, taking meeting minutes, and financial analysis can be accomplished with some success using the early machine learning tools we are beginning to see. Evidence to date suggests that significant productivity benefits will be realized as the technology develops.
But contrary to the marketing rhetoric we're bombarded with every day, adopting an AI solution isn't as simple as signing up for a new SaaS product and setting off into a new world. Before you can even consider an AI solution, you need to address existing issues with your infrastructure and operational processes.
In a conversation with Tech Wire AsiaRichard Davies of OutSystems spoke about the current state of the organizations he interacts with in his work and their experiences with modernization.
Image source: Shutterstock
Foundational Work
While there's certainly a lot of buzz about the potential of AI, Richard suggested that the claim that all companies are moving towards an AI-driven future is largely untrue: “There is some adoption of AI, but I would say it's still very low. If I were to stick my finger up and estimate, I'd say it's at 100% adoption.” [numbers of] Someone who did something [with AI] To be half serious, I'd say maybe 10%.”
Clearly, every enterprise has many tasks that can be at least automated or enhanced by AI, but the obstacles that exist are the same ones that prevent consistent data structures, which are essential for rapid development of enterprise applications, whether they leverage machine learning or not. When software is easy to build (as is possible with low-code platforms like OutSystems), access to partial data is only partially effective.
According to Richard, the core data needed is:[…] The key context you need is the data stored in your core systems, like customer data and core product data. […]. if you want [applications] To really get specific, you need to give them that specific information and ask them specifically what it is that you want.”
Bumps in the road
The roadblocks to gaining visibility into core information include a host of technical issues, including disparate data sources, legacy platforms that don't play well with other platforms, incompatible data formats, and information in formats that are difficult to parse. Additionally, from the business side, there is a perception of risk associated with opening up data silos for ingestion (in the case of machine learning) or use in applications that automate and bring value. For example, the first stage of implementing an OutSystems low-code platform is always to address these issues. But doing the necessary work and addressing these challenges will pave the way for future uses of AI, at least beyond simple, smart customer chatbots. Building AI-powered applications quickly is possible just as quickly as any other type of application, but every application relies on a consistent database.
AI with Low Code
OutSystems customers can build and use AI applications for tasks within their own business areas, but Richard said a secondary area where the technology is being used is in the enterprise application development process that takes place within OutSystems' low-code environment.
In-house software developers have low-code in their range of tools, and AI will be one of them: “So it's like a productivity tool and probably the biggest benefit is that it generates a full-stack app using text prompts and then you can modify it.” Business apps and development tools that use AI are both undoubtedly useful, but as Richard sees it, the majority of the interest is in applications that businesses can build and use to solve specific problems.
“Let's be honest, most people are more interested in business software. 'I need this specific thing for my business.' [app] It's more like “to look at this particular use case” rather than “OutSystems lets you build apps with a text prompt, which is really handy,” which is actually pretty useful, but not that relevant to a lot of people. [of our] “client.”
Image source: Shutterstock
Next steps
Assuming an organization can streamline its data resources and make them a valuable asset, machine learning technology undoubtedly has great potential. Companies can already select and switch between large language models available and use their inferences.
The next stage is to add relevant local data and prioritize the business case for solutions that facilitate automation, applications, or manual processing. Whether the intended outcome is a complete AI-first operational change or equipping your in-house software development team with the low-code tools they need to rapidly iterate on critical enterprise apps, it remains essential to carefully plan technology initiatives in the context of your broader business needs.
To learn more about low-code development, the steps required to create a valuable data fabric, and how to develop the next generation of enterprise-grade software using AI, contact your local OutSystems representative or schedule a demo here.
