Follow Datanami:
March 6, 2019

An Alexa for Your Business


Consumers across the globe have gotten into the habit of asking chatbots like Alexa or Siri simple questions, like “When will it stop raining?” and “Who will start game one for the Padres in this year’s World Series?” Now that level of consumer chatbot experience is coming to the enterprise, enabling employees to get questions to answers about business data.

There are dozens of companies building chatbots for the enterprise, and a handful more that create development kits that let companies build and train their own custom chatbots. Vendors like, TextIt,, and have created their own chatbots to handle mundane tasks.

Amazon is in the business with Alexa for Business, which is designed to help employees manage their schedule and to-do lists, while Microsoft has Azure Bot Service, which hooks into communication channels, such as Cortana, Skype, Slack, Facebook Messenger. Salesforce‘s Einstein bot, meanwhile, is an expert at navigating the hosted CRM app, while Infor‘s Coleman AI can pull answer from some of that company’s ERP systems.

Enterprise Chatting

Another vendor aiming for chatbot supremacy is The St. Louis, Missouri company’s software is looking to carve itself a niche in the middle of the enterprise, where it can become an expert in answering basic questions from data stored across systems, such as the communication channels mentioned above, Microsoft Sharepoint, Salesforce, and Workday.

“It’s like Siri or Alexa for your workplace,” says Dave Costenaro, the head of AI R&D for “We’re an AI platform that connects to all of a company’s data, tools, and information, and make it accessible through a conversational interface.”

Users can interact with through Slack or Skype, or through a dedicated app. A user can ask for information such a phone number or email address of a particular connection, or to pull up a file stored in one of the company’s data silos.


The core of is a search engine that tracks keywords and metadata in the various repositories. Atop that base it uses AI techniques, including recurrent neural networks (RNNs) and natural language processing (NLP) algorithms, to build a more detailed representation of the company’s data, including access patterns. Finally, it employs human “co-pilots” (administrators in the user company) to help organize information more precisely using a graph database structure.

“We use a lot of the NLP techniques that the AI sector is turning out to plug into applications like Workday, Salesforce, CRM, Jira and ticketing,” Costenaro says. “We plug into those and return information, like PDF and Word documents, websites, and FAQ pages. Then we have a human in the loop system to scrounge up and collect data that’s just floating around in people’s heads.”

When has been deployed atop a company’s IT assets (either on-premise or in the cloud), it will enable employees to ask ad hoc questions, such as “Who is the contact in Salesforce for the Boeing account,” and the software will fetch you the data and return it as a text-based answer.

The software is designed to be smart enough to proactively respond to events. According to Costenaro, if the software detects a run on 401(k)-related information or healthcare enrollment information, it can send out an automated note reminding employees of deadlines, he says. “That’s how we add value beyond just being a Sharepoint search,” he says.

The software’s cross-industry and cross-product applicability is also another strength. In addition to hooking into common cloud platforms like Office 365 and Zenefits, the company integrates into industry-specific applications, such as ones used in the mortgage, credit union, higher education, and recruiting industries.

“Cortana is going to be really robust and built out for Microsoft products, and Einstein is going to be really robust for Salesforce products,” Costenaro says. “But we hope to be built out and robust across all of these products and be able to talk to all of them.”

Bots ‘A Plenty

We’ve become accustomed to hearing people interact with smart phone-based chatbots like Siri and Alexa. And eventually we’ll get used to hearing employees requesting things from chatbots and robots, Costenaro says. hooks into a variety of enterprise systems and data repositories

“I think over time, as voice becomes less of a weird thing, we’ll have people talking to computers and robots in the office,” he says. “We can add voice-to-text later on pretty easily.  We can do that today with the mobile app.”

For now, most interactions will be silent and typed manually onto the screen, but eventually they will become verbal. Whether we’re all talking at robots or writing them questions, get used to having bots in the workplace – most likely lots of them, from many different providers, Costenaro says.

“I don’t think that anyone is going to solve all of this. There’s going to be an ecosystem of bots and apps and people and groups and webpages,” he says. “But the closer we can get to a single solution, if we can eliminate some of the clutter in people’s lives and help them to do their work better – I think that’s going to make them happier and engaged and waste less time. That’s the goal that we’re striving for.”

Don’t DIY

Much of the technology needed to build an enterprise or consumer chatbot is open source that’s freely available on the Web. But Costenaro warns do-it-yourselfers against trying to build something themselves. integrates with real estate systems

“The devil is in the details in putting all this stuff together,” he tells Datanami. “If you click on one of these algorithms, it works pretty good, just in the same way that a Sharepoint search works pretty good. It’s going to get you what you need in the top five blue links maybe 50% to 75% of the time.

“Any one of these algorithms that Facebook or Google put out probably does the same right out of the box,” he continues. “So that’s not really going to move the ball for you. It’s not going to give you the insights or the chat interface.”

It’s a big leap to go from a series of unconnected technology projects to a full-fledged enterprise chatbot system. Costenaro says the biggest advantages that a firm like has are its engineering tenacity and its ability to iterate atop multiple customers data.

“Where we gain a competitive advantage is ensembling [the algorithms] together and covering the edge cases and having a tight iterative loop with product and design,” he says. “It’s all about having the data, having the experience of working with the customer, honing and fine tuning this stuff. You’re not going to get a professional experience just building it yourself with off the shelf component. That’s the evolution that we had to go through in prototyping and building out NVPs. was founded in early 2017 and has attracted a handful of customers, including Washington University in St. Louis, USA Mortgage, and Ameren. The company raised $8.4 million in a Series A last year.

Related Items:

The 5 Fundamentals of Bot Deployment

Bots Are Learning Their ABCs, But They’ve Got a Ways To Go

AI-Powered Bots Gearing Up to Serve You