CST: AI platform for creating text and voice robots

The main articles are:

• Speech Recognition (Technology, Market)
• Speech technology: On the path from recognition to understanding

2024: GigaChat Integration

The group of companies CST integrated in their own AI platform for creating text voice and robots LLM neural network is a model, and GigaChat has also improved the dialog analytics module. The updated AI platform will make it faster and easier to create dialog assistants. The solution will be especially relevant for large businesses and will public sectors help automate internal processes, improve customer service in. contact centers The CST announced this on October 25, 2024.

The AI platform of the CST has received a whole range of updates. The built-in RAG algorithm allows you not to write all scripts for the robot manually, but to connect large amounts of data and generate responses to the user from an external knowledge base, which saves up to 70% of the work on developing the script. Dialog assistants will be able to answer almost all customer questions, since they are no longer limited to a narrow scenario and differ in empathy - now they are much easier to adapt to the audience, to choose a communication style. This brings the quality of dialogue assistants to a fundamentally new level, allows businesses to increase the economic effect and create humanocentric services. New features - the foundation for the future development of the platform as a multi-agent solution that will allow you to organize a whole system of skills within the framework of a single AI assistant, using various external data sources and LLM capabilities, noted Dmitry Dyrmovsky, CEO of the CST group of companies.

Among the key advantages of the platform is a built-in analytical module that allows you to compare dialogs with customers by many parameters: time slices (before or after the launch of a new service or product, service), by channels (allocate a voice or text channel), by business parameters (analysis of different categories of customers - by regions, with a connected service, VIP) and dozens of other metrics. Such reports provide the business with a visual picture of the effectiveness of the chatbot, depending on the specific conditions.

The introduction of semantic document search algorithms has increased search accuracy in an unstructured database by up to 90%. So the introduction of the updated AI platform will allow you to provide more complete answers to user questions, including complex ones, while the client does not receive unnecessary information, but only the necessary one. The added platform capabilities are effective not only in dialog assistant scenarios for customers, but also in the co-pilot assistant mode for internal employees and operators.

Added the ability to adapt the image chat boat for a specific client in real time, directly during the dialogue. The user can customize the stylization of the robot's responses and tone in the script, depending on the context of the dialog or client parameters, and different formulations do not need to be set manually in the script. The robot script uses one answer, but before a response is issued to the client, it goes through a "stylization" stage: polite, detailed, formal answers - for a mature audience, more concise and less formal - for a young audience.

The CST AI platform combines the ChatNavigator omnichannel dialogue platform, which is responsible for developing and executing scripts, and VoiceKey.IVR, a software component that provides voice interaction with dialog assistants and IVR systems with speech synthesis and recognition functions.