AI and automation

As an ILO report on GenAI and jobs put it: "The chat function dramatically shortened the distance between AI and the end user, simultaneously providing a platform for a wide range of custom-made applications and innovations." From predictive maintenance in manufacturing to chatbots in customer service, AI-driven solutions and smart automation had already transformed industries. However, GenAI's newly unleashed accessibility and accelerant sophistication ignited our imagination. Moving forward, we knew that combining an openminded, enterprising attitude about AI with our secure, compliant-by-design work ethic could lead to game-changing gains for our customers and ourselves.

At Schuberg Philis, automating as much as possible is a long-standing principle applied not because we want to replace people with robots. Rather, we want to free our colleagues and our customers' colleagues from repetitive, uninteresting tasks. In its place, we encourage them to engage in more interesting work and, in the first place, to have time to be inspired and inspire others to do so. While AI has been projected to cause job disruption, the risk of job loss due to occupational automation has subsided and given way to AI's potential for work augmentation, the ILO report points out. Respondents to PWC's Global Workforce Hopes and Fears Survey 2023 similarly reported AI's positive impacts more than negative, with the most often cited being by 31%, who expect "AI will help me increase my productivity/efficiency at work." Meanwhile, technology leaders already identify AI and machine learning "as the most important technology for achieving short-term ambitions," notes KPMG's Privacy in the new world of AI report.

Emergent technologies are also proving to be a valuable component of security, one of the cornerstones in all our solutions. As findings from the IBM Cost of a Data Breach Report 2023 show, "even a limited effort to integrate security AI and automation into security workflows can offer a significant acceleration in the time to identify and contain a breach as well as a sizable reduction in costs." What's more, organizations that used these capabilities extensively within their approach experienced, on average, a 108-day shorter time to identify and contain the breach. They also reported 1.68 million euros' worth in lower data breach costs compared to organizations that didn’t use security AI and automation capabilities.

Positive impacts, powerful automation

While GenAI's development has revolutionized how the world creates content, behind the scenes, it has also impacted how engineers work. Its capacity to handle enormous amounts of data quickly and smartly has made room for engineers to take on a more curatorial role. At Schuberg Philis, this means that if we use AI-produced content as input, we keep the human in the loop to thoughtfully curate outcomes in accordance with our standards. Investments we are making to deploy large language models (LLMs) to sbp.cloud, our private cloud, are enabling us to better understand how these models work and can thus be optimally applied in our solutions. This past year, we also built chat.sbp.ai, our own safe environment allowing our engineers to use GenAI. The user interface that we created permits Schuberg Philis users to select any of the LLMs that the company currently offers for assistance with coding for software development. In the meantime, to further optimize their efficacy, we are training these LLM and have enabled different locations to host them depending on data classifications.

Both for individual users and within the context of large-scale enterprises, AI is enhancing decision-making and operational efficiency. Ideally, generative AI can let us further “shift left," a virtue we uphold because addressing concerns early in a project's life cycle generates efficiency and speed, keeping our customers ahead of the competition. For example, when writing software for a bank, we know that our code must be highly secure and run a lot of automated vulnerability detection. AI-aided programming can offer immediate feedback on the quality of the code as we write it, for instance, by pointing out potential security flaws and preventing logging of personally identifiable information (PII).

However, AI and automation are applicable to more than our own work performance. We believe AI holds tremendous potential for engineering the complex problems that our customers frequently face, and have already used it for mission-critical solutions. In one case, we applied machine learning and computer vision to detect human activity in the prevention and detection of falls among nursing home residents. For another customer with hypercompliance needs, we created a composable AI platform that enables customized speech-to-text note-making, saving hours of time on tedious reporting. For yet another, we built a chatbot that helps employees quickly extract information from their organization's huge body of documentation saved across disparate platforms.

Ethical and responsible use

Although we appreciate AI as an exemplar of our innovative age, we also acknowledge it as just one among many tools. The tools we select, furthermore, are simply those that we know will deliver 100% impact for our customers' business. If and when we use AI, it is not for the sake of exploiting a new technology, but rather because its deployment makes the best business sense. The value tracking framework we apply is therefore focused on capturing the tangible benefits of our AI initiatives. Using data extraction and surveys, we seek to gain early-stage insights into usage and user satisfaction. This allows us to learn and improve quickly, tracking patterns and best practices to guide further development and, eventually, to scale within those areas that generate the most value.

In a similarly pragmatic spirit, we subject AI to the same scrutiny as all our other technologies, services, and ways of working. It must serve to guarantee uptime while being secure, compliant and – considering how new this frontier is – held to the highest standards for ethical and responsible use. Earlier in 2023, Schuberg Philis drafted company-wide guidelines to help us simultaneously keep up with AI and regulate our use of it. Our AI manifesto, as we call it, walks a fine line between urging colleagues to practice risk management and mitigation while encouraging creativity. To establish the manifesto, we identified best practices in ethical use of technology that were relevant to AI, traced how they link to our values as a company, and gathered input from colleagues of varying disciplines and viewpoints. This exercise yielded 10 principles intended to ensure that our approach to AI stays meaningful, impactful, effective, and ethical.

Evaluating the ethics, security, and privacy implications of AI is a practice that comes naturally to Schuberg Philis. Abiding by the principle of tech for good means that our company believes technology can serve a better world. Doing so, however, demands that we continuously assess how our choices – technological or otherwise – will affect people and the planet. As such, we have a team dedicated to exploring how our colleagues and customers can tap the potential of AI and smart automation all while making a positive impact on society. And because we know there is still much to discover in and around AI, we plan to invest more knowledge, skills, and solutions toward its successful deployment.

Optimized relevance, optimized value

2023 provided an opportunity to build the Schuberg Philis value proposition on AI. Much like the field itself, our proposition continues to be finetuned, to deepen and to broaden, as do the parameters that define it. Above all, we are convinced that our exceptionally high-quality engineering applied in combination with AI and smart automation's unprecedented power can converge to optimize relevance – and value – for our customers. Ranging from software and data to digital resilience and cloud, our service offerings can each benefit from having possible pain points and gain points viewed through the lens of AI. At the same time, our expertise in these focus areas informs our capacity to best leverage AI.

This past year, we also began working on AI development projects for banking and financial services and, crucial for this sector, testing ways to increase reasoning in AI. So far, our use cases span an array of solutions entailing, for example, creating AI-based interaction with end users, building LLM-based natural language processing (NLP) applications, helping organizations with the everlasting challenge of processing large chunks of unstructured data for legal and technical compliance, threat modelling, and content management. Customers whose data must stay on a sovereign cloud can take advantage of sbp.cloud to run particular AI services. Regardless, whether via workshops, concept sprints, process blueprinting, or any other method that puts our AI and smart automations to test, we will continue exploring how to best deliver business progress while pushing the boundaries of intelligence, both artificial and human.