To close (minimize) this chat widget, click on the (X) in the right upper corner

phrase2vec is a sequence embedding algorithm our company has come up with. It's like word2vec, but for multiple-word questions and commands. It maps short phrases to vectors in 300-dimensional vector space, so as semantically similar phrases cluster together. May I have email to send more information to?

Phraser API can help you to build your Alexa skills, Actions on Google, etc. It is a semantic classifier for natural language questions and commands. It converts fuzzy unpredictable user utterances to formal structured commands. Using the latter, you can then write your code in any programming language or framework you like.

You don't need any dialog management tools! Write your program as if there is no natural language input. Write your normal deterministic code in any language or framework you like. Our Phraser API will take care of unpredictability of natural language input.

Time and Date API extracts time and dates from free text; returns them in a structured form (e.g. “what if we go there next Sunday morning and return between Nov 11 and 13”). It is more comprehensive and accurate than most other solutions. Pre-trained – no further training required. May I have email to send more information to?

Destination recommendation engine can recommend leisure travel destinations by activities or themes – think “Pandora for travel” (e.g. “7 day vacation for two in unspoiled nature, with good biking”). Available via an API or at http://alterra.cc

"Garbage" special class is for completely useless queries. Garbage. Trash. Spam. When you encounter it add it to the training corpus; it will help ML to learn to recognize it. ...

"Ignore" special class is for meaningful queries worth answering that however shall not be added to your FAQ file. They may be off-topic, one-off, too ambiguous, too short, too long and complex for the bot, etc. You may forward them to humans. More here

"To do" special class is for meaningful on-topic questions that don't have answers in the current FAQ file, but should. You may create new articles and then re-assign these queries to them. Thus, the "to-do" name. More here

Currently, the system supports only English. Other languages can be added on demand for large accounts. It is relatively straightforward: our system doesn't rely on linguistics, sentence parsing rules, etc. It's 100% Deep Learning. We just have to update word embeddings.