Building the Intelligent Enterprise Powered by AI
Building the Intelligent Enterprise Powered by AI
By Gareth Martin, Chief Executive Officer
The advances in AI, and particularly in agentic AI, present an opportunity to rethink how we optimise our businesses.
The value of process re-engineering and automation have been lauded for many years. The value of AI based innovations is the hottest topic of the decade. The combination of the two is where the true value opportunity exists. That value can be realised locally within individual processes of the organisation, but the multiplier effect of optimising across, and between, the processes and functions of a business is where the real value lies.
This opportunity to optimise the whole system, the functions and their processes that make up a business is made possible by the latest AI capabilities being deployed in conjunction with leading edge architectural approaches.
Stock Replenishment Example
Stock Replenishment Example
Consider the replenishment of stock of grocery products in a distribution centre which serves e-commerce customers as well as physical stores. Within the replenishment process there is significant automation with human validation points as well as some manual data entry. The process uses many complex data processes and calculations(represented in the diagram to the right).
These processes work together to replenish the stock. The more effective and accurate the process the better the revenue, the lower the waste, and the lower the cost of the process.
When we reconsider the complex process above as a decision flow we can imagine how many of the decisions in the process could be optimised, automated and streamlined, and also how the impact of accuracy in calculations across the processes can impact costs or revenues. This gives us the motivation to optimise the business function and consider how AI can support that.
How to Optimise Complex Systems
How to Optimise Complex Systems
The requirement to build intelligent, automated and optimised processes, business functions and systems demands a technical solution approach that is specifically designed to address the following four key needs.
1.Intelligence
In any process or system, whether it is simple or complex, intelligence is required at various points to enable decision making and task selection and execution. Traditionally intelligence comes from a combination of analytical, problem-solving and decision making capabilities delivered by humans using data (and technology). To support acceleration and scaling, the decision making and action taking needs to be shared with intelligent machines. This has been possible for some time for simple tasks using automated deterministic functions. For more complex tasks requiring reasoning we add AI (GenAI) components (agents) to the network to replace or augment the steps traditionally taken by humans.
2.Automation
Automation of the end-to-end processes above with the additional element (as opposed to traditional approaches) of agency given to the reasoning elements (the GenAI components) gives us the opportunity to significantly increase the proportion of the system or process which is automated and limits the need for manual, human intervention, focusing that intervention on critical control points or strategic direction setting. The greater the portion of the overall system or process which can be automated, the greater the speed of execution of the process and therefore the efficiency of the system from a time-to-completion perspective.
3.Optimisation
Optimisation of the system as a whole is covered in automation, above, however there is also the consideration of the mathematical optimisation of specific elements within the system. When we consider intelligence we do not limit ourselves to reasoning but also consider the intelligence associated with complex mathematical and statistical functions. This is why machine learning and statistical modelling functions (can be considered non-reasoning agents) make up a critical part of the overall optimisation of complex systems. The ability of the GenAI elements to trigger these functions, consume and understand their outputs, and use them in their reasoning, adds significant power to the overall capability.
4.Governance
In building the capabilities above, and integrating them into a connected network of intelligent processes, automatically calling each other to support completion of their own process objective in the context of the overall system objective, and in doing that leveraging new technologies like LLM's, consideration for the governance and security risk needs to be top of mind. In this context we consider governance from the perspective of monitoring, managing, securing, auditing and controlling the intelligent system.
Rethink, Re-use, Recycle
Rethink, Re-use, Recycle
In order to achieve true scalable success and value realisation in the coming transformation, and get unstuck from the frequent focus on localised use-cases and pilots, we need to follow a strategy that has an end goal that is both ambitious and achievable with iterative value realisation and that avoids over reliance on individual technologies.
Rethink
Firstly we need to rethink (or perhaps more realistically reimagine) how our business operates. View the operation as a complex system of business functions and processes. All of which are inter-connected. Sharing information and triggering other processes in different parts of the organisation. Consider the value of information at critical decision points and the potential for bottlenecks or uninformed decisions across the entire system. Now imagine you could optimise it. Provide the relevant information at the point of decision making to a human or machine that could make a more informed decision, more quickly, triggering the relevant processes associated with that decision in real time, disseminating instructions across the business and observing the impact of that decision, also in real time. It's your business but accelerated, informed, super-charged.
Re-use
Affordable scalability is critical in programmatic modernisation of the business, as such one of the key concepts to abide by is re-use. Architectural patterns and re-usable templates reduce the overall costs and make every additional process easier and quicker to optimise. Those patterns cover both functional and non-functional parts of the solution, but also extend to the principles of interoperability and extensibility. Intelligent agents across the business should have the ability to request access to information or execute processes in other systems if it is in line with their goal achievement and does not contravene security or safety protocols. This prevents the need to replicate the same or very similar process in an additional system just because the data sits there.
Recycle
Maintenance, repair and upgrading of the system prevents accumulation of technical debt and reduces long term risk associated with aging technical solutions. Designing a modular, partially microservices-based, system means that components can be readily swapped out or upgraded in isolation of other parts of the system. API contract based communication makes it easier and quicker to automatically test the impact of component replacement or code changes, whether those changes or replacements are developed manually or automatically.
Where to Start - Finding the Value
Where to Start - Finding the Value
It is important to start in the right place. There is often a temptation to start somewhere "safe" but our recommendation would be to start in the area that has the fastest route to significant value, and to be safe in your execution rather than playing it safe in your selection.
Find the Value
It's surprising how often businesses struggle to calculate and articulate the financial value associated with the projects they want to embark on. For a programmatic evolution of the business creating a consistent approach to do this will make a significant impact on the approval process. A good place to start in finding the high value areas is to follow a methodology such as value-chain-mapping. This will assist in the identification of the processes and decision points that have the largest impact the top or bottom line financials. These are the best places to start as the optimisation of these elements will generate a financial return to support continuation to subsequent optimisations.
Assess the Readiness
Consideration of readiness from the perspective of people, process, business cycle, data and technology is important. In assessing the readiness of the business to implement and realise value from the optimisations, you will identify the challenges that need to be addressed for each process before it can be optimised. Start these readiness activities early on and in parallel so that they can be removed as barriers.
Select Strategically
The compounding effect of optimising multiple, interconnected processes along the value chain is greater than the sum of the individual parts. However, optimising one part of the overall system in isolation may result in a bottleneck. When mapping the value chain it is important to understand the inter-dependencies in the system and the knock-on impact of optimising parts of that system and not others. Design a roadmap which will enable a compounding, step-wise realisation of value.
