Problem:
You have a tech support team and want to answer some common questions automatically, or just want to lower the support team load.

Solution:
We build a chatbot, which classifies incoming tickets, responds when it can using state-of-the-art technologies and neural networks (such as BERT) and switches the dialog to human operator only if it is necessary.

Problem:
You want to know what people are thinking about your company.

Solution:
We can build a scraper of news sites or social networks and build ML models, which can provide a sentiment score for any particular message or for the entire people segment. Also, we can develop a web application, where you can explore the results in real time.

Problem:
You have a delivery company and wish to find ways to improve your business to optimize problem solving for your customers.

Solution:
We download your users' feedback and extract meaningful problems from the collection of texts, provide specific examples supporting our findings and suggest ways your business might be improved.

Problem:
You have a tech support window on the website and receive huge amount of spam messages. Or, you have an e-commerce website, and people are leaving fake reviews for your products or create fake listings (they don't have a real product, but upload stock photos and copy-paste product descriptions from the internet) and cheat others.

Solution:
We build a system that detects such malicious user behavior in real time with reasonable precision and recall.

Based on text descriptions within real estate sale listings, mine data to locate only those properties that interest the buyer (near the subway, quiet place, high ceiling) and update database in real time.