Digital Transformation: the role of Artificial Intelligence on automation initiatives

How does AI impact the automation of administrative processes?

etermax tech
etermax technology
3 min readOct 6, 2022

--

By Fernando Mogetta, Head of AI Consulting, etermax AI Labs

The COVID-19 pandemic accelerated the transformation processes of companies, and, in many cases, changed the priorities of digital initiatives. This is the case of the automation of administrative processes, which became a primary priority simply because people were no longer physically in the office to carry out these processes manually.

According to a study carried out by Forrester Research, companies reprioritized those initiatives that were indispensable for the operation of their business under the new conditions (resilience zone), such as business continuity and bots based on AI for customer and employee support, together with initiatives focused on cost efficiency and reduction (acceleration zone), such as the automation of administrative processes.

Document processing is a great example of this. In highly regulated industries, such as Financial Services, Insurance, Utilities, and Telecommunications, thousands of physical and digital documents are processed daily (client balance sheets and articles of association, legal documents, agreements, new regulations, among others). These documents involve repetitive work done by qualified personnel. One of the documents some banks have started to automate are official acceptances of credentials.

Artificial intelligence, combining its abilities to ‘see’ (Computer Vision), ‘read’ (Optical Character Recognition), and ‘understand’ (Natural Language Understanding), allows for these documents to be automatically processed. It extracts all relevant information from each of them, structures it, and makes it available to the people or apps that need it. Among its main benefits are:

  • Increase of efficiency and productivity (cost and time reduction).
  • Acceleration of the client onboarding process.
  • Fully digital processes, capable of absorbing sudden increases of volume (scaleable).
  • Increased satisfaction of collaborators, as time destined to routine tasks is made available for tasks that have more added value for the business and that are more defying.

How is this achieved? By combining different automation tools with AI abilities, such as BPM (Business Process Management) and RPA (Robotics Process Automation), to automate a document’s full lifecycle. Firstly, Computer Vision can be used during the pre-processing stage to improve image quality. Then, OCR (Optical Character Recognition) can be used to extract the text from the document, and NLU (Natural Language Understanding) specialized processing models can be employed to identify and extract relevant parts of each document. Finally, structured data is organized in a database that can be consulted directly or with an API. Depending on the use case, a revision stage of the data identified with a trust level below the minimum threshold may be included.

These projects require coordinated work done by a team usually following agile methodologies and made up of business users, computational linguists and data scientists, who will apply NLP models to each problem, and IT developers and architects who will implement these models and the automation of the full process.

A recurrent lesson learned by companies who have worked on these kinds of projects is that, from the point of view of architecture, it should be presented as a services platform with reusable components that can be reused to implement multiple Use Cases that require NLP abilities.

Leading companies in the adoption of these technologies report accuracy levels of AI models above 90%, reducing the total processing time of documents up to 90% and increasing employee productivity by 70%.

At etermax AI Labs, we have achieved similar results in the implementation of process automation projects which allowed us to contribute to the potential of Artificial Technology in our clients’ processes of Digital Transformation.

--

--