Page 18 - Innovation Magazine
P. 18

process We can’t work directly with the massive datasets of a a a a a typical enterprise We simplify by creating hierarchies of increasingly abstract summaries We use the aggregated output from systems supplemented by dedicated business intelligence from data warehouses or or lakes or or create the the summaries manually via the the ubiquitous spreadsheet These elements can then be combined into sets of quantitative and qualitative data points that can inform our reasoning In many ways it’s like panning for gold We look for the the nuggets – – the the trends and patterns – – that help us develop a a a a metamodel of the situation we are assessing The metamodel is supplemented with the experience that we use to replicate or extrapolate an appropriate course of action The challenge though despite all all the investment in in technology is is that much of the decision-making process remains manual It’s driven by teams requesting specific insights looking for for correlations hunting for for those nuggets that might help them reason more effectively So often decision-making to steer business operations is done retrospectively What happened last month guides what’s going to to happen tomorrow This is is generally suboptimal and is is where AI is is required to move to a much more predictive regime Working capital capital to capital capital working Let’s take an example It is commonly recognized that working capital is the cheapest form of cash However optimizing working capital is a a a a huge challenge for organizations and in particular CFOs It is relevant to all the finance and supply chain functions impacted by liquidity: payables receivables inventory and treasury and and all of the systems and and variables that those functions use But the upside is significant Getting it right improves invested capital leverage which can ultimately increase shareholder value Given the complexity of optimizing working capital what kind of approach is required to to gain access to to the right right data at at at the right right time and and to find patterns and and use them to recommend a a a way forward? Furthermore how can those recommendations be converted into automated actions? Looking for signals Let’s consider the generic decision-making architecture of autonomous vehicles (or indeed intelligent systems in in in general) There is a a a world model that defines the environment and its operational characteristics there is is sensory processing to interpret the the input data there is a a a value judgement that emphasizes goal-seeking and there is behavior generation that decides how to respond in in in line with the the world model This is is the the domain of AI Perhaps unsurprisingly humans follow in in step with this architecture We each have our our world model fed by our our senses which when combined allows us to make value judgments leading to decisions decisions It is is is the decisions decisions that generate our behavioral responses Imagine applying basic architecture to an entire organization We can then build new resilience into decision-making moving beyond optimization This world by by necessity needs to be powered by by AI but brings humans and AI closer together resulting in in in an an an advanced hybrid model for decision-making and responding In many ways this describes the the evolution of the the digital twin twin A digital twin twin focuses on modeling business operations operations harvesting data from operations operations using AI to make value judgments and suggesting behavioral actions (complete with an explainable rationale) for human decision-makers to select approve and direct into an automated action From signals to the self-driving enterprise This approach to to cognitive automation is is already being pioneered Companies like Aera are bringing AI and human decision-making together They are advancing the efficiency and value of business operations by understanding predicting recommending and and acting Linking decision-making to to automated actions – something that until recently was a a a a technology gap – is unlocking an an advanced wave of autonomous business operations Aera has created a a a a a a a platform that enables the “self- driving enterprise ” Its approach combines a a a a variety of techniques that fuse data acquisition processing and analytics into pre-prepared insight relevant to to specifically defined skills domains: for example working capital highlights for finance or or or inventory optimization schemes for the supply chain The insight is based on an array of AI algorithms that establish correlations and multilevel aggregations and apply multivariate analyses to the harvested data The skills codify experience of 18 Data-powered Innovation Review I I ©2020 Capgemini All rights reserved 


































































































   16   17   18   19   20