Context-Aware Automation: Learning the what before the how

Automation is here to stay

Over the course of the past decade and across industries, the most aggressive adopters of automation have been banking, financial services, insurance and health care. Document processing, specifically, has been an important focus area for most organizations supported by both regulatory tailwinds — most notably the transition away from Libor — and the business-as-usual disruptions caused by the Covid-19 pandemic.

Image by Canstock

IDP solutions that exist today have been able to deliver several tangible benefits primarily across:

  • Efficiency, by replacing thousands of hours of manual effort creating meaningful cost savings;
  • Risk reduction, by eliminating human error to reduce risk in operational processes; and
  • Turnaround, by reducing turnaround times.

Gaps in today’s solutions

Several practical challenges stand in the way of automation’s ability to deliver on its promise. These are:

  • Inconsistency: The processing of documents that do not adhere to a standardized format, such as contracts or reports, cannot be easily automated. This also extends to documents created by third parties, where document form or structure can change with no prior notice.
  • Complexity: Automation also fails to work on complex documents. These include documents with a combination of text, tables and diagrams; documents with a blend of landscape and portrait pages; and documents with multiple languages.
  • Extensibility: Traditional approaches to automation focus on the “how” by recording mouse clicks or keystrokes. This results in the automation process relying primarily on formatting cues to achieve its objectives, without real understanding of the underlying context, limiting the possibilities of extending the automation beyond the original use case. 

Learning the what before the how

Human subject matter experts learn to solve processing problems by taking a “what” approach. They do not start with the task at hand but instead spend time learning the underlying context. Once sufficiently aware of the context, they use format or structure in a document, only insomuch as it allows them to abstract away key information relevant to the completion of the task.

Automation processes can emulate this approach by abstracting and persisting meta-data around previously completed tasks & documents. Knowledge Graphs are one possible means of achieving context abstraction and retention. 

The promise of context-aware intelligent automation

Financial services firms receive tens of thousands of documents every day from vendors, customers and counterparties from diverse data sources across email (inline and attachment content), fax, snail mail, APIs, web and mobile applications.

In many of these instances the format, structure and complexity of the documents vary significantly over time and across sending parties, even though the context remains exactly the same. This is especially true when it comes to free-text-heavy documents, such as contracts or annual reports. It is these documents that test the limit of traditional automation today and which context-aware automation promises to automate. Examples, to name a few, include:

  • Scoring a listed company on the basis of annual reports, sustainability reports, press releases and other documents on ESG metrics;
  • Account openings, loan approvals and claims processing, in situations where information that is present in a form needs to be validated against documentary evidence that has been submitted; and
  • Due diligence of third parties across vendors, acquisition targets, counterparties and customers. 

The path forward

As automation becomes mainstream and business leaders start to focus on automation as a key lever to enhance the top line, rather than a means of cost or risk reduction. The resulting focus from short-term problem solving to long-term value creation will see the development and deployment of context-aware intelligence platforms. Developing these platforms will require patience and capital, and it is quite likely that in the near term they will fail to do substantially better than traditional solutions.

However, in the longer term, these platforms will become integral to the future of work, forming the base for the next generation of knowledge workers to develop, deploy and derive value from versatile and extensible automation solutions.

Prashant Vijay, CEO of Romulus, a document intelligence software provider

A veteran of the financial services industry, Prashant Vijay is currently chief executive at Romulus, which specializes in building software products that automate document-heavy operations in the financial services industry. He has spent more than two decades working at the intersection of technology and data across multiple roles and geographies. His views are informed by his experience in tech and business roles at Goldman Sachs, and his sales and product and business management roles at IHS Markit.