Imagine if software was invisible but intelligent; imagine if you could ask that software a question and get an insightful response; and imagine you could interrogate your corporate data using that invisible software without ever needing to hire a data scientist.
That’s exactly what Hyper Anna did.
The Sydney start up, led by co-founders Natalie Nguyen, Sam Zheng and Kent Tian, has developed a machine intelligence layer that sits on top of a collection of data. Nguyen explains; “Our vision is for people to be able to interact with data by just asking questions using natural language like ‘how’s my sales doing’ and the machine intelligence agent just goes away and looks at the data and analyses that and brings back insights for you.”
While the software itself may be almost invisible to the end user, its impact isn’t. Hyper Anna is at its core a machine intelligent agent that performs the role of a data scientist. If you had access to a data scientist (at a cost of around $150,000 a year at the current going rate) you could ask them a question and they’d get back to you in four or five hours.
According to c-founder Nguyen, when Hyper Anna goes live in early 2017, people will be able to talk to the system, send an email or text a question and get an answer in three seconds for around $99 per user per month. And, unlike human data scientists, Hyper Anna is available 24×7.
And is it really invisible?
“It’s software you don’t see most of the time. There are three ways to interact with Anna, you can either go into a website and basically the data been piped directly to Anna’s brain and sits in an Azure cloud, then you can ask a question in the web site. Or you can cc Anna in an email conversation, so say for example if you email me saying there is a talk at 4pm today about your sales performance and you cc Anna in that conversation, Anna can read from the email context and understand that you will need some information about my sales performance and then email that information. The third way that you can interact is just text a number, just like texting a person, and Hyper Anna will be able to get back to you.”
Because of the machine learning technology that underpins the system, the insights delivered are sculpted to reflect the real intent of the question.
“When humans ask a question we always have an intention for the question. If you have asked me ‘how are my sales doing?’ and I told you that you had sold $100,000 of product, that is actually not very insightful at all. Your intention is to know whether your performance last week was really good or was it not good and why,” says Nguyen.
By mapping the possible intent of a question, monitoring how the response to that question is received and used, and what follow up questions are asked, the machine intelligence agent learns which information has been most useful and ensures that informs future responses.
Hosted on Azure, Hyper Anna has access to any data that the user wishes. Often that means that corporate data is piped daily or weekly into Azure where it can be accessed, while any software as a service systems such as Salesforce and Xero are accessed directly via open APIs.
To help an organisation decide what data Hyper Anna will need to provide users with valuable insights Nguyen says she asks them first to tell her the ten questions that people might ask – the questions that would help them do their jobs better. With those questions framed, it’s generally clear what data sources they will need to provide access to, she adds.
Leading Australian insurance business IAG is currently working with a beta version of Hyper Anna on top of one its data sources. End users – who could be relationship managers or sales managers, who may be unfamiliar with Excel and SQL, and certainly aren’t data scientists can still ask a question of Hyper Anna in natural language and get the insight out straight away. HyperAnna is also working with Westpac on a project.
To support the growth ambitions of the company, which include a US launch in 2017, Hyper Anna has completed a seed funding round securing $1 million from Westpac’s venture capital business Reinventure and is finalising a further $250,000 investment from AirTree.
“Building a machine intelligence tool for the workplace is one of the really big topics in tech start-ups at the moment and will fundamentally change the workplace in 5-10 years’ time,” says Nguyen.
“For us the secret sauce is that our team are all data scientists – it’s different from a tech start up teams where the majority of people come from engineering backgrounds. Our approach is data analytics for everyone – not just the select few,” she adds.
Hosted on Microsoft Azure, Hyper Anna is testing a series of Azure services including Service Fabric and Data Lake to see how they could enhance the solution.
Being cloud based was a no-brainer for the company according to Nguyen; “The brain of Anna constantly needs to reap feeds of data so it gets smarter – so having a central place for the brain to sit is extremely important and also helps with deployment time compared to an on premise solution. We deploy literally within a day.”
