AI and automation

About the company

Artificial Intelligence in communication and customer servicing is a set of methods and technologies for solving a broad variety of tasks. It allows to reduce service and labor costs significantly, while simultaneously improving the quality of service.

Our Expertise for various AI technological domains

SoftBCom constantly assesses the universum of AI-related solutions and seeks out the best options to integrate with. We run feasibility analysis for possible application of such solutions in different industries, evaluating our customer`s potential for efficiency gains with AI. And in the end, technologies employed depend on a given project’s requirements.

We at SoftBCom have experts proficient with the following domains of AI:

Machine Learning (ML)

Deep Learning

Natural Language Understanding (including LLM)

AI in Customer Service - From Potential Analysis to Proof of Concept

With our structured process model, we ensure that AI is used in your customer service center where the cost-benefit ratio is right.

In the AI potential analysis, our experts for AI in customer service illuminate your processes along the typical fields of application for market-ready AI solutions:

Virtual Agents:

Voice / chat bots for case-closing transaction processing in self-service.

Life Agent Support:

Context-sensitive provision of knowledge data, work aids

Interaction analysis (voice, text, image)

After "triggers" for the application of prefabricated responses and automated workflows.

During the Proof-of-Concept phase, we create pilot installations in prioritized application scenarios to evaluate acceptance and benefits. Thanks to our Cloud SaaS operation and the integration of AI partner solutions into our Contact Center solution, pilot projects are supported flexibly and with minimal effort.

Implementation areas / cases

Here’s a list of implementation areas that we have studied closely (either a production or pilot installations are in use):

1. Categorization of requests

This is the most popular usecase, which provides the fast return on investments, saving human labor and increasing the accuracy of categorization in comparison with manual operation. AI categorization is applicable in SoftBCom Contact Center and Service Desk for voice requests, chats, or e-mails . It also can be used in Service Desk operations for attributing registered tickets or incoming e-mails.

2. Multi-step dialogues through different channels

There are often dialogues that resume after some time, and sometimes even through other channels. In each session of such a dialogue, the client receives the required response. However, Chat GPT (and other LLMs) does not remember conversations. We're building a bridge. We add a framework to make LLMs work in functional business context. The information from this client’s previous requests is put into unified history of interaction, which increases the accuracy of answers and the degree of client satisfaction.

3. Extracting structured data from customer requests

Often clients’ requests are a large piece of unstructured information. E.g. a customer is calling, but all the the agent needs at this stage in the conversation is the name and order number. Extracting such useful data and providing it to the agent in condensed structured state is a task for AI.

• For text messages (for example, emails) AI recognizes and extracts the necessary data items and attributes them to the elements of the structure. E.g., the agent receives name, request order and problem type.
• In the case of voice communications, the speech is initially transcribed (converted to text) and then analyzed according to the procedure described above.

4. End-to-end processing of requests

Certain repetitive tasks do not require involvement of human agents, but can be easily be delivered in an AI-powered customer self-service set-up. Here are some examples of usecases:

• Find out the status of the order (at what stage of processing, date of delivery, etc.)
• Coordination with customer to determine the date and time interval (for transportation, delivery, servicing, doctor's appointment, etc.)
• Collecting meter readings (electricity, gas, water, etc.)
• For transportation companies, providing information on the schedule and cost of tickets for the right time and direction
• Providing detailed personalized advice on the terms of a loan, an insured event or the procedure for self-serviced maintenance
• Selection advice and information about the availability of goods in an interactive form

5. Biometrics

Biometrics (voice or image) can be used to identify customers. For customer identification by voice, a voice channel is sufficient, for image identification a video call is required. Using these means, customer identification can be fully automated.

6. Speech analytics

Speech analytics of agents is carried out for control of customer service quality and for additional agent training. Speech analytics of clients allows the agent to quickly adjust to the mood of a problematic customer or request help. In addition, analytics allows to extract the necessary data elements (for example, keywords).

General implementation principles

Best of breed solutions.

In our projects at any specific moment we rely on the best existing solutions in the market that would best fit the customer’s task, and integrate them into our system. We also offer some solutions of our own development. Here are some AI solutions that we use to incorporate into our systems. SoftBCom continously monitors the market for new AI solutions and cooperates with partners after thorough feasibility analysis. For all of the solutions we have pilot installations and ready to use integrations, as well as modules or connectors:

01

Google Dialogflow

AI Chat and Voice Bot with NLU

02

Open AI - GPT 3.5 (“Chat GPT”) and GPT 4

Generative AI Voice Bot with NLU based on a Large Language AI Model

03

SoftBCom ML Subsystem

Text request categorization

04

Spitch set of solutions

Chat Bot, Text Bot, voice analytics, voice biometry

05

Vox Implant

Smart connector for Dialogflow and contact center (voice channel)

Data security

One of the key questions for using a given solution is the importance of data security. All the public cloud solutions have potential leakage problem. When data security is crucial, we can use on-premise solutions. When that is impossible, we can go for SaaS with an anonymizer.

Advantages of different platforms:

Cloud: Price-effective to deploy and operate

On-premise / private cloud: Full control over personal data access and storage

We provide all SoftBCom solutions on all platforms:

Cloud

Private cloud

On-premise

Possible escalation for best CX

AI helps a great deal in solving repetitive tasks and lifting this weight off human shoulders. In fact, it does such tasks better than humans! Depending on customer’s business processes, AI can help the agent to solve requests quickly and effectively with categorization, voice biometry and data extraction. That means better and faster answers and better customer experience. End-to-end processing has the same goal – it is available 24/7!

AI for better CX

• AI is better than humans in categorization, voice biometry and data extraction
• End-to-end processing is available 24/7

Escalation for bespoke service

• When needed, AI can be blended with live agent servicing
• Provided with SoftBCom Service Desk and Contact Center platforms

Depending on our customers’ business processes, end-to-end processing can be blended with live agent interaction. Complicated situations still could require human involvement. In such cases we integrate live agents into request processing with our Service Desk and Contact Center platforms to run the whole process seamlessly, at minimal labor costs and to best client experience.

Want to know how SoftBCom can improve your customer support business processes? Sign up for a free consultation