MindMeld's Deep-Domain Conversational AI platform marks the next stage in the evolution of conversational artificial intelligence (AI), said the company's founder and CEO Tim Tuttle.
Released on Friday, the platform makes it possible for companies to create voice and chat assistants that can demonstrate knowledge and expertise around any custom content domain.
The platform offers capabilities such as broad vocabulary natural language understanding, question answering across any knowledge graph, dialogue management and dialogue state tracking, and large scale training data generation and management, with cloud-based or on-premises deployment.
The AI assistants can also be launched across a variety of interfaces, such as the company's native application or website, to third-party devices and applications such as Amazon Echo, Facebook Messenger, Google Home, Microsoft Cortana, and Skype.
Tuttle said it typically takes a few weeks to get a voice or chat assistant up and running. However, when it comes to licensing fees, it varies between companies based on the complexity of what needs to be built and the volume of end users.
"Once these applications are live, [customers] can see real user traffic and the machine learning algorithms get smarter over time with use," said Tuttle. "With millions and millions of interactions, they become bizsmart, so it generally takes several months if not a few years to unlock the true value of this technology."
The San Francisco-based company claimed that the reason previous assistants have been underwhelming is because open source or academic AI toolkits such as TensorFlow, CNTK, SyntaxNet, CoreNLP, and NLTK provide little data suitable for production-quality applications; and that cloud-based developer tools such as Facebook's wit.ai, Google's api.ai, Microsoft's LUIS, and Samsung's Viv provide limited support for building AI models for custom knowledge domains.
The company was founded five years ago while Tuttle was doing AI research at MIT, and it took the company about three and a half years to build out its core technology, with the first version of the platform launched about a year and a half ago.
"Our original investors invested in the company so that we could build this machine learning technology that would finally understand human language," Tuttle said.
The company has raised about $15.4 million to date from a consortium of investors including Google, Samsung, Intel, Telefonica, Liberty Global, IDG, USAA, In-Q-Tel, and others.
It has also acquired some high-profile customers including music streaming service Spotify and most recently global fashion retailer Uniqlo.
MindMeld's technology is seeing traction across the board, from ecommerce and travel to food ordering and consumer banking, according to Tuttle, who noted that it coinicides with the rising enterprise interest in conversational AI. He added that the company has experienced about five-times growth in the last 12 months.
The last year has been big for the emergent conversational AI market.
Amazon made headlines for its AI personal assistant chatbot Alexa, which is embedded in its voice-activated bluetooth speaker device Echo. Alexa is able to answer questions, report the weather, set alarms, add items to a shopping list, and order a taxi. It's been reported that more than 4 million Echos have been sold thus far.
In September, LinkedIn launched new bots to help people make the connections and build the skills they need to advance their careers.
Companies like Google, Facebook, Amazon, and Microsoft launched new developer APIs for their messaging applications, virtual assistant platforms, and voice-enabled devices.
Google's acquisition of api.ai, a California-based startup that provides a platform for developers to build chatbots, coincided with the launch of Allo, a messaging app for Android and iOS with smart features like the "smart reply", which suggests replies to texts and images based on the context of the conversation.
The Microsoft Bot Framework, launched earlier this year, recently surpassed Facebook in popularity, with more than 45,000 developers using it to create bots and machine-learning programs for Skype, Facebook Messenger, Kik, and other chat platforms. This is compared to roughly 34,000 developers making bots on Facebook messenger.
Latest statistics show Facebook Messenger also has about 34,000 bots, with the company also announcing analytics support for bot developers.
In September, at the Oracle OpenWorld conference in San Francisco, CTO Larry Ellison showcased Oracle's new chatbot development platform.