How artificial intelligence dramatically increases the productivity of the back office
Problem of productivity
Each of us wants high-quality and fast service both in the commercial and government enterprises/ But increase in staff for faster service leads to additional costs, a high load on the HR department, and higher prices for customers. What is often unacceptable
Creating more with the same efforts is a key task for quality service and is one of the modern challenges for both the state and business.
The problem of scaling
From time to time there are cases when the workload on employees increases dramatically. Such a challenge, for example, was the epidemic of Covid-19, when most of the issues needed to be resolved online, an additional load on medical and delivery services was created. An additional burden on migration services was caused by the crisis of illegal migrants from Belarus to Lithuania, Latvia and Poland.
Hiring and training additional employees is a rather large investment of time. The same employees have to train, who are already under increased workload. At the same time, when the crisis has passed, the recruited employees, the hiring and training of which took time, effort and money, will have to be fired.
Financial decisions are made on the basis of figures collected in tables, based on calculations. It's not that simple with text data. It is not an easy task to quickly understand what topic the last 3000 requests were on, what dynamics were in the topic of requests over the last year, whether the employee mistook the topic of the request. But this information is critical for making strategically correct decisions and quality control of the services provided.
In this article, we will describe how we solved these problems using one of the artificial intelligence methods - Natural Language Processing (NLP).
NLP is the solution to many problems
Natural Language Processing (NLP) refers to the branch of artificial intelligence or AI concerned with giving computers the ability to understand text and spoken words in much the same way human beings can. Thus, NLP allows the computer to adequately respond to users, make extracts from the text, classify requests and appeals, analyze texts and speech as a human would do, but thousands of times faster.
For example, in one of the projects that we carried out, a fairly large number of messages came to the city administration. There were many repeated messages that got to different employees and each of them spent time searching and solving the same problem. After the implementation of our system, all incoming messages before reaching the person were processed by an algorithm in order to identify repeated calls and prompt the employee how to answer the best.
This made it possible to speed up employee responses by 60%, devote more time to complex issues, thus increasing customer satisfaction. For some of the most frequent types of questions, it was possible to automatically provide information, almost completely relieving the burden on staff.
Then we connected the recognition of toponyms and supplemented the clients' appeal with data from external systems. Thus, employees did not have to spend additional time searching for information from the organization's systems, they received it along with the request and responded to requests faster. Reducing the time spent on one call was another 20%.
In another project, along with text data, requests within the company were sent in the form of document scans. It took employees additional time to open the document, compare the correctness of the data in the letter and documents, and retype additional information from the letter.
Our algorithm for finding the desired text on a scan of a document and its recognition allowed us to simplify the work by 1.5 times and reduce the number of errors by 60%. Such algorithms are indispensable when it is necessary to check or reconcile documents along with a message from the user.
Also, we performed analytics by type of messages. Valuable information was obtained on what problems prevail in different cities of the region. Where there are more questions about medical care, where about education, where about housing. This data made it possible to more effectively distribute efforts and ensure quality control of the services provided.
The advantage of using artificial intelligence techniques for text processing is simple scalability in case of increased workload. There is no need to hire additional employees, it is enough to increase the server capacity, and this is much faster. As a rule, explosive growth occurs in questions of the same type, and it is easier to automate their answers.
In this small overview, we have shown how NLP ensures:
- 1) Improving the efficiency of back office employees through automatic recommendations for an answer based on previous answers.
- 2) Helps to process text and graphic data, reconcile information from different sources.
- 3) Supplements queries with data from related systems.
- 4) Provides high-quality analytics based on text data.
If your organization works with information in text form, you have a support service or citizen appeal, your processes can be improved. Call us and we will be happy to tell you how.