Skip to content
Augmented Intelligence Certification

Smart Process Automation – useful conversation


Intelligent automation is a well-known issue today. The technologies all around which it is constructed – like augmented intelligence certification, robotic procedure automation, all-natural language processing (NLP), and cognitive computing – are regular organization matters.

Nonetheless, these spots need to not be observed in isolation from a single yet another. They need to form aspect of a whole. What is extra, they require context – and that context is the incredibly useful just one of the corporation in just which they are to be applied, and the processes to which they will be used.

At Capgemini, this holistic perspective has been termed Smart Procedure Automation. In this piece, I’m going to glimpse at two factors of Smart Procedure Automation with which people are most possible to have interaction, involving technologies that permit human-computer conversation.

Voice assistants

Voice assistants are potentially the most obvious occasion. They have two principal uses.

The initial and at the moment the most widespread software is to manage inbound calls. Their use minimizes stress on shopper assistance teams, by enabling technological innovation possibly to help you save some time, by inquiring for and getting typical details such as account details from buyers or to save a good offer of time, by dealing with full client interactions. Intelligence created into the system that recognizes phrases this kind of as “I want to converse to an agent” can help phone calls to be switched at any issue to a shopper service representative for exception managing.

This is what the baseline technological know-how can realize, but the total positive aspects can only be recognized when this intelligent automation is constructed completely into processes. At Capgemini, we have been doing work tough to outline and produce interfaces with significant enterprise platforms such as SAP, AWS, and Salesforce to enable voice assistants to be customized to specific corporate needs.

The next principal application is the handling of outbound phone calls. In this application, AI voice assistants can get in touch with buyers to supply suggestions, for instance, or get in touch with delivery motorists to present updates when they are en route.

As you could expect, a lot of development get the job done is essential in this location. For instance, at a regulatory degree in quite a few nations, outbound voice engineering desires inbuilt security to confirm the id of the man or woman answering the connect with and at a useful amount, we’re performing to combine outbound voice into enterprise methods such as MRP and offer chain administration options to raise its applicability.

All round, the efficacy of voice assistant technology and in certain, its intelligence, will be the end result of steady enhancement. The much more iterations there are – the extra sample discussions and knowledge voice assistants can muster – the better ready the technological know-how will be not just at comprehension the human voice, but at decoding imperfect input, these kinds of as speech from which terms are lacking, or unclear, or seemingly out of context.

In this discipline in typical as properly as in particular regions which include finance and health care, we’re doing work with the greatest practitioners in the marketplace, doing the job via normal shopper situations, identifying pain points, and creating responses.

Text recognition – and beyond

Speech has a corollary in crafting, not just as a indicates of interaction, but in conditions of human-computer system interaction.

Just as voices range in pitch, tempo, tone, accent, and far more, so handwriting can differ in a terrific variety of ways, and for machines to understand it, optical character recognition (OCR) devices will need to be built-in with intelligent algorithms and subjected to a terrific deal of apply ahead of deep learning strategies can start to bear fruit.

It is not just about handwriting, both. Unstructured information in typical – which as properly as handwriting consists of movie, audio, graphic data, and PDFs – is developing at a phenomenal fee. In 2017, IDC forecast that by 2021, at minimum 50% of international GDP will be digitized.[1] Which is a substantial total of facts, and Smart Course of action Automation will have to have to be place to operate on substantially of it – for instance, looking at and responding to a handwritten observe from a purchaser who wishes to cancel a subscription, or assessing and processing scanned promises varieties.

A situation in position

A number of of these smart human-pc interactions can be seen at get the job done in a cognitive assistant we not long ago designed for a client. The aim was to boost the speed and lessen the price of a funds assortment procedure, whilst sustaining the sense of human engagement – even even though the functionality was now automatic.

Our AI-primarily based client option contacts consumers who owe cash, generating either courtesy phone calls or telling them payment is overdue. It has been implemented making use of cognitive NLP, voice transcription, the cloud, microservices, and various modern day world wide web frameworks, and it supports 24 languages, together with much less common kinds this sort of as Dutch and Finnish.

The resolution has a semantic awareness of people’s responses, and notes their promises of…