In a world of manual processes, conflicting rules and decisions for the same outcomes, it has proved almost impossible to standardise the rules and options required for automating processes.
In most large enterprises there are conflicting / duplicated processes and rules within those enterprises’ business units or departments, many of which are not (or fully) documented and never get to a stage where all processes and rules are standardised throughout the entire enterprise.
Mapping and standardising processes and data artefacts to enable automation, efficiency and cost savings are the ideal, but can take years to achieve and may never fully reveal overlaps, gaps and inefficiencies.
Rationalisation and standardisation exercises are often done manually, using business process management (BPM) tools to see overlaps, duplications and gaps. However, the logic / decisions taken along these processes are usually hidden within applications and configurations, and it is almost impossible to unearth them in order to properly examine the process landscape and decision patterns.
Compounding the challenges, BPM projects to achieve organisational process standardisation and automation are notoriously difficult and time-consuming, particularly in large enterprises with many business units, branches and departments.
The AI solution
Machine learning (ML) is a subset of artificial intelligence (AI) and presents a solution. Given the metadata of an organisation’s data and process landscape (and/or the information from its BPM repository), AI and ML can alleviate the process mapping burden for employees and help data (or “meta”) scientists to apply algorithms to find patterns, decisions and options, as well as any conflicts, overlaps and gaps. This lends itself to metadata analytics (MA) but with AI speeding up the mapping and unearthing exercise.
While these technologies are still in their infancy for use in process standardisation, we already see interrogative algorithms traversing and reporting on metadata. To support process standardisation and automation, the AI simply needs to be applied in new ways to identify decision points, options taken and gaps in the processes.
While these technologies are still in their infancy for use in process standardisation, we already see interrogative algorithms traversing and reporting on metadata.
From a data-driven approach, the application of AI on an organisation’s historical data and current “end-point” metrics / results will reveal decision patterns and evidence of options taken along the organisation’s processes.
For example, to reveal the standard metric for tallying the number of sales per day in a retail chain, the AI might indicate the metric is based on the sales transactions data, less VAT. However, the SMEs on the ground might say they use multiple data sources and calculate an average to determine the number of sales per day. So the findings of the AI analysis can be reconciled / compared to that of the SMEs – this is a great way of “unearthing” undocumented SME information or knowledge.
By using AI to search historical trends in data that has been generated and stored, as well as on endpoints, the organisation can achieve a high level view of patterns. This then provides a blueprint of evidence that forces people on the floor – the decision-makers, option-takers and/or SMEs – to fill in the gaps, refine the model, provide evidence of their actions along processes or take decisions based on the blueprint. This approach can dramatically simplify and speed up the process of achieving standardisation and automation.
Even if options and decisions taken are manual, un-documented or not configured in system processes, examining the data and end results will enable the analyst to identify decision / option patterns.
In contrast with the months or years it would typically take to unearth all relevant information around processes, data and rules, the scanning / exploration of historical data and metric (end results) by AI is possible in a fraction of the time.
Given the objectives / process or data domains / use cases, AI applied over the data and respective metadata might take only a few days or weeks to trawl all of the organisation’s historical data, find all trends and patterns, look at the metrics and link the metrics back to the data. If processes are already documented in BPM tools or in metadata, the AI results can be linked to give a view of the process landscape in terms of process duplications, overlaps, gaps, decision points and options.
This then, also addresses technical insights on the operational back-end, and insights on decisions. Organisations would be enabled to identify differences in rules or timings, find commonalities and zoom into conflicts and problems.
By having these valuable patterns mapped, organisations can use the AI learnings and input this data to robotic process automation and business intelligence analytical tools for leveraging further value and supporting both efficiencies and governance.
Share