Microfinance Institutes have a noble reason for living: by servicing people without access to the regular financial system, they aim to grow financial inclusion. We learned earlier this year, in the Afterwork session with Accenture and microStart, that the impact of this additional inclusion has an exponential network effect on society “where entrepreneurs ignite others to more creativity and exponential employment, as these entrepreneurs need more employees to help the company grow”.

UiPath invited me to a hands-on webinar recently, organised by the Microfinance Center, about the importance of artificial intelligence in Microfinance Institutes (MFI). That got me thinking: why should they need the power of AI more than any other financial institution, and what could be the challenge for them to embrace that power?

The value of hyperautomation

The resources in MFIs to deal with the administrative burden are scarce, and requests are high. So is the cost for processing them, compared to the ticket size of microloans and the higher exception handling in the credit scoring process.

This is where the power of automation and AI comes in. Technologies like the UiPath Automation Platform demonstrate incredible value to help these MFIs to do more with less and to meet business needs more quickly.

During the Microfinance Center webinar, Nitin Purwar, Senior Functional Architect at UiPath, explained the three different automation technologies that, when mixed together, deliver what Gartner calls hyperautomation:

• Robotic Process Automation (RPA): emulates humans and focuses on execution. RPA is very much rule-based and extremely useful for repetitive tasks of structured data. (See also our blog on “Automation is making banking more human again”)
• Machine Learning (ML): will assist humans and focuses on thinking. ML is capable of working on more complex tasks with both semi-structured and unstructured data. It is capable of recognising patterns, but it requires a lot of training.
• Document Understanding (DU): can either be rule-based or model-based and will help the organisation in dealing with all kind of documents and transforming it into structured data usable in the context of RPA, for example.

MFIs that embrace hyperautomation to deal with new loan applications, customer onboarding and all kinds of validations will save time. These time savings enable their employees and volunteers to bring real value to their members and customers by guiding and improving the day-to-day business, tasks where human contact makes a real difference.

Nitin: “The UiPath Enterprise Automation Platform enables organisations to easily discover automation opportunities, to build automation from the simple to complex, leveraging sophisticated AI, and manage them all with enterprise-scale orchestration.”

MFIs and the power of hyperautomation

Nitin: “With client onboarding, for example, there is obviously a lot of paperwork required with plenty of validations and screenings before you can service your customers. All this is traditionally pretty manual, especially for MFIs since they deal almost always with small-scale customers.”

Hyperautomation helps streamline the entire process for onboarding, which, in turn, brings down the onboarding time.

The audience for MFIs is very diverse and sometimes more fraud sensitive than people with access to regular banking. Combining RPA with machine learning allows the robots to learn how to validate forms and documents, and they often see fraudulent patterns that the human eye would overlook. Aside from being an incredible time-saver, this reduces fraud and the resulting reputational damage and post-processing.

Another example: UiPath cut the processing time for an SME lending application from 45 minutes to 2–3 minutes per loan. That increases a company’s capacity with 1800%. Or, to put it differently: it frees up 95% of the application time to create value for the entrepreneur instead of assisting him in his administration.

Finally, there are complaints and general requests for the contact centre. Hyperautomation will help with the classification and categorisation on the one hand and resolving the claims whenever possible on the other.

According to Nitin, hyperautomation can help MFIs to answer up to 30%-40% of different types of customer requests and complaints. This is the cost saver on top of the time saved by automatically reallocating the remaining incoming requests to the right employee to answer them.

Challenges to embrace the power of AI

The idea of ‘well-trained robots’ brings us to the essence of the challenge for MFIs to embrace the power of AI: the required training. And training requires time and dedication of people in the field.

Training also requires clean data, a lot of clean data, to train the different models where ML is implemented.

We all know the saying: “sh*t in, sh*t out”, so we need to prepare data as well, to make sure the organisation starts with a clean sheet. Each scenario requires the right data sets for a period of a minimum of 2 months as a starting point, according to Nitin.

Luckily, companies like UiPath have the right capabilities to assist the organisation in training the different models and to make sure the training can start with clean data.

A second essential challenge comes down to trust and the capacity to hand over your decision-making to a robot. What level of accuracy are you willing to accept without human intervention? Some cases will be easier to hand over than others. Repetitive questions in a chat or through social media will be much easier to delegate to a ‘virtual employee’ than, for example, the validation to a foreign identity document.

Machines learn and keep learning, so accuracy will go up over time. Until then, the process may require temporary additional resources. They will support the training, evaluate the process for further improvement and make sure you don’t leave your sleep in the beginning while risking wrong decision-taking.

Resources are scarce, but keep in mind the end-goal: over time, hyperautomation will bring down the human interventions in the impacted processes. It is an investment, but an investment worth making, which over time will result in more business and happier, better-served customers.

Read the original article here.