Quality speaks for itself
The strength of advanced conversational AI lies in its ability to be used for multimodal purposes. It enables chatbots to answer queries faster and more accurately, empowers virtual assistants to recognize various voices in noisy environments, and converts text to natural, human-like speech across a wide range of languages.
Built upon a foundation of state-of-the-art Natural Language Understanding technologies, AIQ.TALK Chatbot is an AI-based builder for businesses and developers who seek to deploy a virtual customer service. No matter the industry or size of your company, AIQ.TALK Chatbot works for any business to deploy rich consumer interaction and to optimize operational efficiency. When it comes to implementation AIQ.TALK Chatbot provides the full scope of options included SaaS, private cloud, and on-premise model.
High precision and recall accuracy
With our hybrid approach of rule and deep learning based intent classification, AIQ.TALK Chatbot demonstrates higher accuracy compared to other engines, especially in the Korean language. Our tool allows you to build sophisticated chatbots easier and more efficiently. Only a small amount of training data and a small effort to register sentence sequences are needed.
Natural non-sequential conversation
AIQ.TALK Chatbot’s Dialog Manager can analyze complex and non-sequential conversations to identify the most appropriate responses.
See entire conversation flows at a glance and manage them efficiently with our chatbot builder. Our award-winning UX/UI enables developers and non-experts alike to easily create sophisticated chatbots.
Optimized integration and collaboration
Multiple users can simultaneously train the data and manage the chatbot system. Our bulk testing tool for system automation and the dialog scenario test tool support RESTful API for a seamless linkage to external systems.
What makes AIQ.TALK Chatbot different?
AIQ.TALK Chatbot comes with a rich set of features to enable easy use for everyone.
Whether you are a developer, product manager or customer service specialist, we have you covered.
|a||Other existing global solutions||AIQ.TALK Chatbot|
|AI system automation test||Not available||Intent recognition performance test system,|
dialog scenario test tool
|Multiple user management||Partially available||Fully available|
|Dialog switching and context management||Only one-way switching||Available|
|Logs & statistics dashboard||Partially available||Accumulative trained data for each version|
|Data export||Only for JSON||Supported XLSX (Excel) and JSON|
AIQ.TALK MRC (Machine Reading Comprehension) enables machines to understand a text and answer questions in a given context, using the latest deep neural language model. Businesses can deploy automatic question answering service through AIQ.TALK MRC without further data training.
Proven performance in accuracy
AIQ.TALK MRC boasts the highest answering accuracy, superior to human performance and other AI language models in Korea. Our machine reading comprehension is ranked first on KorQuAD (The Korean Question Answering Dataset) dataset, of which task is to predict the best answer from question and context pair.
Ask and answer like chatbot
Accelerate building a chatbot for answering questions from customers. Just by uploading desired documents and feeding them AIQ.TALK MRC, companies can easily build an intelligent chatbot that is able to answer from given documents with minimum resources.
Automatic customer screening via phone call
When combined with AIQ.TALK STT, AIQ.TALK MRC can provide answers in “Yes” or “No” to customer’s questions during a phone call by finding the exact questions and answers in a given document. The combined solution can immediately handle tasks such as screening or testing customer’s qualification.
Rapid search on video / audio clips
Transcribing audio from text in real-time, AIQ.TALK MRC finds the answer to the question and plays the exact clip where the answer is included. Design a novel service beyond the bounds of keyword search, based on our neural search engine that provides the answer to comprehensive questions within a video clip.
Core technology: NLU (Natural Language Understanding)
Skelter Labs’ NLU technology uses machine learning to reveal the structure and meaning of text. It extracts relevant information, thereby illuminating the context of a conversation. Powered by our high-performance NLU technology, the trained model can answer specific user questions, classify content, support virtual assistants, and more.
Best-in-class performance in both Korean and English, even when given complex intents or a small amount of training data.
Entity Recognition & Slot Filling
Increase accuracy for fat-head words as well as expand coverage for long-tail words with a hybrid approach incorporating a comprehensive Knowledge Graph and models for Named Entity Recognition (NER).
Machine Reading Comprehension & Question Answering
Automatic comprehension of questions in text and conversations. In-depth analysis of the meanings of words are used in a query to suggest phrases for answering with high precision and recall.
Efficient management for both sequential and complex non-sequential conversation flows with advanced dialog switching.
Our speech-to-text product, AIQ.TALK STT converts voices to text by applying most advanced machine learning techniques.
High performance in noisy environments
Easily understands voices, especially in low-quality audio environments such as phone calls. Automatically transcribes audio in real-time, understanding context to recognize even mispronounced words and non-standard dialects.
Powerful speaker diarization and recognition
Strong performance on voice classifications even with populated surroundings and multiple simultaneous speakers. The to-be-recognized voice is extracted from the environment and converted automatically into text.
Various pre-built models
Choose and apply from a selection of pre-built models for different use cases including mobile, PC, phone calls and video.
Only available in Korean
Our text-to-speech product, AIQ.TALK TTS reads the structure of sentences to generate human-sounding speech.
Natural speech with a small amount of data
Understands and extracts the distinct pronunciation and speech patterns of a speaker in order to generate natural speech.
Generates and synthesizes speech only with a small amount of training data by applying advanced deep learning.
Faster conversion speed
Powered with an advanced speech synthesis algorithm, the speed of converting text into speech is faster than that of other solutions. Increased efficiency by converting maximum amount of texts seamlessly and rapidly.
Customization & tuning
Customize SSML, speaking rate, pitch, and volume depending on the use case.
Only available in Korean