Today’s businesses must process an astonishing amount of data to make decisions, even in operational areas such as supply chain, procurement and contract pricing. Assessing these extraordinary volumes of data is nearly impossible for even experienced managers.
Through distributed computing and powerful artificial intelligence (AI) and machine learning (ML), decision process automation (DPA) can address this complexity by automating how organizations analyze massive volumes of information. Simultaneously, human experience and insight complement algorithmic decision-making by assessing contexts too nuanced for AI and ML to address as of yet.
DPA offers a new way to optimize decision-making by automating how organizations analyze what has happened and what is happening to more accurately predict what could happen.
Technology has created enormous challenges by growing the amount of data businesses need in order to compete — but technology also offers the solution. Organizations can continuously model, simulate and select the best choices for decision-makers — automatically — by using distributed computing, algorithms that can process massive amounts of data at lightning speed, and so-called evolutionary AI.
This is the opportunity that DPA presents, the logical next step from robotic process automation (RPA) and intelligent process automation (IPA) in the evolving continuum of business process automation (BPA).
RPA improves operational efficiency while reducing human error. IPA helps organizations automate complex tasks, leveraging cloud technologies to bind together operational systems. And other functional software addresses challenges in supply chains and procurement. All are effective, because they allow knowledge workers to focus on more value-added tasks.
But none of these operational technologies can aggregate multivariate data; simulate and model the effects of different decisions past or present; or rank-order optimal decisions based on desired outcomes, allowing managers to then choose one.
DPA can. It leverages massive computing power and increasingly sophisticated algorithms to assist in business process decision-making. Just as one might create a “digital twin” to model performance parameters on an aircraft engine, DPA creates a digital model of an organization. It examines historical data, trends and decisions; evaluates current conditions by analyzing incoming data streams; and models potential scenarios and outcomes. Using DPA, organizations can better examine evolving opportunities and formulate previously unseen solutions. Call it a look into potential futures.
Moreover, DPA can capture and retain the elements of inherently loose processes, such as individual decisions or tacit processes, by extracting and tracking key process details. And by analyzing a simulated instance of an organization’s past choices and current data inputs, DPA optimizes decision-making in situations where even the best minds may balk or fall short. It winnows out optimal choices from overwhelming amounts of data, incorporating algorithms that progressively refine outcomes while eliminating human biases.
DPA delivers choices based on how the organization has worked in the past even as it incorporates data streams from the present. This brings humans back into the loop precisely at the point where the human capacity for insight and foresight, and the ability to understand institutional context, offer the most value.
In short, we believe DPA is the future, because it shows organizations what the future can look like — machines present optimal choices, while humans remain in the decision loop.
How DPA works
DPA does not eliminate RPA or IPA, but uses data generated from both as input. Fundamentally, it is a higher-order decision engine running in a virtual model of a business’s operations. It relies on evolutionary AI, running iterative sequences of scenarios, with thousands or even millions of variables, on a cloud-based network of millions of distributed processors.
This sequential algorithmic analysis allows the decision engine to rapidly refine to optimal outcomes, enabling automated decision-making at lightning speed. DPA makes choices as humans might, but with the ability to manage an extraordinary number of complex variables, including data inputs from evolving conditions in real time, without overlooking data or making decisions based on hunches.
Consider how proactive insight and rapid scenario modeling could save millions of dollars in contract pricing annually by repricing contracts for key commodities, raw materials or parts as conditions change. DPA could automatically optimize the structure and terms of deals and streamline customer interactions. By contrast, consider how an individual (or even a team of analysts) often struggle to optimize decision-making in rapidly evolving circumstances. The figure below shows how DPA can use information streams from tens of thousands of customers to predict order patterns across products.
DPA gives enterprises new, more flexible decision support — the power to evaluate constraints and prescribe the right spend to the right customer at the right time across every channel, automatically.
Driving better decisions
DPA offers organizations a deeper view of operations and provides more clarity in resolving complex problems and predicting outcomes. It provides a framework to help companies create higher value — a decision-optimization capability that supports the organization’s ability to:
Analyze and correlate historical data, details on outcomes of earlier decisions, and incoming data from current activities in order to anticipate future outcomes.
Adapt to changing situations, such as evaluating contract performance, comparing the relative cost of spot purchases, examining trading and/or hedging decisions, managing supply-chain risk during weather events, or reducing logistics bottlenecks and costs even in a crisis such as the COVID-19 pandemic.
Gain insight into what is happening in near real-time, discovering the causes of certain outcomes and adjusting proactively to evolving circumstances, while eliminating human biases.
In the context of supply chains, for example, DPA ML algorithms could perform scenario modeling on demand, based on inputs about changing conditions. This could lower inventory requirements while reducing unfilled sales and the number of changes to sales orders. DPA could also optimize delivery volumes, sales routes and vendor selection, increasing on-time delivery while smoothing disruptions. And the technology could intelligently improve contract pricing, reducing uncertainty in changing financial conditions and increasing profits per contract.
DPA can build rules based on correlations drawn among the conditions present when a set of decisions were made, factoring in their downstream impact. Iterative modeling toward outcomes means that once a path to an outcome is explored, a business can feed process output data continuously back into the DPA engine. Applied on a larger scale, it can generate insights about the outcomes of historical decisions and their results, which human decision-makers can then use.
Evaluating the timing of implementation
We have pointed out to many businesses that they need not revamp their RPA and IPA efforts to adopt DPA. Indeed, information delivered from automated processes at a lower level is vital to driving the DPA strategy. Bringing DPA online necessitates understanding specific needs for decision-making and where it would have the greatest impact — to accelerate and optimize organizational responses. The following conditions could signal such needs:
Existing rules or scenarios are used to address certain situations, and they are repeatable.
A combination of events or environment-related issues occurs repeatedly, and automated rules could augment human decision-making.
The business encounters scenarios that never happened before and must conceive rules to address them, while leveraging such new rules to address other scenarios with similar characteristics.
Such conditions suggest asking the following questions:
In what areas could DPA most benefit managers and supervisors? In managing inventory? The supply chain? Freight management and costing, and ensuring on-time delivery? Adjusting contracts or procurement? Revenue collection?
Which key performance indicators (KPIs) correlate to decisions? How could monitoring those KPIs help verify the benefits of DPA? What is the frequency and volume of decisions?
What applications or systems does the business use to make decisions? Do different departments rely only on certain systems to make decisions?
What sources of information or knowledge could contribute to better decision-making? What form of collaboration should be enabled?
Does the business chronicle sets of circumstances and the external conditions of decisions?
At what points does the organization create, capture and process automation data?
With answers to these questions and the input of an advisor experienced in implementing DPA, businesses can create a roadmap to realizing a more agile, accelerated, and higher-value decision-making function.