Applied AI – a gamechanger for business operations

The competitive advantage promised by applied AI is very much here. And it’s delivering competitive advantage for organizations in the transformation business operations.

As my colleague Arul Pradeep writes elsewhere in this edition of Innovation Nation, practical examples are generally better than theory, and show is generally better than tell.

Which is why I thought it would be useful for me to provide examples of applied artificial intelligence (AI). The two implementations I’ve summarized below are very much in the real world.

Language dependency reduction

Whether they are for customer services, supplier assistance, or HR purposes, the support functions of major enterprises have a major hurdle to overcome – and that’s language. Global enterprises must serve the information needs of everyone who gets in touch, no matter where those people are and no matter what language they speak.

Multilingual helpdesk staff can help, but that only gets you so far: they can’t cover every translation permutation. Nor can they give assistance at the scale a multinational organization would need. This is partly because they can only help at a conversational level and can’t translate documents over the phone, for instance.

Capgemini’s language dependency reduction (LDR) solution was developed to meet this challenge. It automates the translation of text in documents while maintaining their format and document structure. This helps reduce dependency on language resources and perform operations globally.

What’s more, it infuses custom translation instructions with enterprise glossaries that extend beyond standard language to improve translation quality. These include generic business terms and acronyms such as “P2P” (procure-to-pay), but also includes context- and abbreviation-aware text enrichment specific to domains and organizations, such as abbreviations of product names and context-based translation.

The solution can handle emails as well as documents in xls, pdf, jpg, doc, html, and ppt formats. It can translate over 150 languages, it holds its data securely, and it easily integrates into applications APIs. It’s this specificity of the solution to the language, terms, and customs of individual enterprises that sets it apart from other online translation offerings.

Predictive analytics

A US-based multinational office supply company was constantly improvising its collection strategy. Before payments had reached a point at which they were deemed to be ageing, it was difficult to go through call logs manually to check whether settlement had been promised or refused. With over 60,000 call logs each month, it was proving to be a challenge to manage the process manually where actions are pending to expedite cash collection.

Capgemini proposed a scalable solution driven by AI and natural language processing (NLP) that can read through the user comments from the call history and automatically derive insights from the call log. It can also classify the call logs into the required groups to derive business insights and provide courses of action for the collection strategy.

Each call is automatically collected from Webcollect and the conversation transcript fed into NLP-based model to provide predictions in real time on the required onward action. The call log is classified and assigned to one of 12 different categories by the solution’s machine learning algorithm. The system also integrates these predictive analytics with an Intelligent Control Center dashboard that provides overall visibility over finance operations.

So far, our client has found the solution’s predictions are 90% accurate. Cash collection has been expedited and efficiency has improved. What’s more, trials have shown not only that the approach is scalable, but that model accuracy can be heightened with the introduction of more data.

Tangible results

At the time of writing, these two projects are close to deployment, and the competitive advantage they promise to deliver explains why I’ve been unable to name the clients involved. It’s safe to say, though, that both organizations are excited by what they’ve seen so far.

Artificial intelligence is very much here, very much now – and it’s very much delivering.

This article is published in the new edition of our Innovation Nation magazine. Read more from our special feature on “Automation and the data-powered organization” and download the full magazine