One of the key take-aways of our data analytics module in my digital innovation class was that the first step to great data analysis was asking the right questions.
Ayasdi works on the premise that “We need better algorithms to ask the right questions”1. That is, people exploiting data may not always know the right questions to ask, and advanced mathematics offer solutions to interpret large amounts of data starting with the data itself.
Ayasdi (“to seek”, in Cherokee) is a Palo Alto startup founded by Gunnar Carlsson, Gurjeet Singh and Harlan Sexton. Gunnar Carlsson is a mathematics professor at Stanford, the two other co-founders did extensive work together with Professor Carlsson. This academic bias is probably the main competitive edge of Ayasdi: the firm is able to investigate mainstream applications of cutting-edge research. Since 2010, the company has raised $106.35m of funding in 7 rounds 2. Pre-2010, millions of government funding contributed to research that made the eventual commercial solutions possible.
The solutions developed rely on topological data analysis. Topology studies geometric properties of data, and TDA looks at the topology of data sets to recognize patterns and analyze the data. The objective is to allow faste As someone with limited background in data science and advanced geometry (but wishing I had a math major right now), I certainly do not understand all of the science before this analytical tool. However, what is clear is that traditional data analysis has always used statistical tools to work with large amounts of data. In this radically different approach, variables represent coordinates of a data point in an n-dimensional set, and geometrical analysis is used to analyze data, and due to this fundamental difference in approach, different analysis and results become possible.
The company provides this graphical example of a TDA application in oncology.
For further reading, Michael Lesnick offers a helpful primer on TDA3, and Prof. Carlsson offers an introduction in this Ayasdi video.
Data analysis at large has become ubiquitous, and it can be hard for many players to differentiate their value offerings, and they often run the risk of clients replicating their offering internally and cutting them off. For Ayasdi, at least today, the value proposition is somewhat clearer as it offers analytical capabilities that most organizations cannot match internally today. As such, value capture is also made easier for the firm, with a more obvious value proposition to customers, charging customers for the service also becomes easier. Indeed, Ayasdi has been able to develop relationships with large players mainly in financial services and healthcare, industries where it offered the first tailored services, and is continuously expanding into new areas. Due to the complexity of its solutions, one element that can be difficult relates to customer education. This may explain the wealth of information that can be accessed on the Ayasdi website. Looking at the resources tab, videos and solution briefs across industries allow better understand how TDA can support companies’ needs. Selected applications include detecting fraud, assisting the drug discovery process, allocating assets in a portfolio, enhancing client experience (in banking).
Clearly, the promise shown by TDA is vast. Ayasdi has generated significant buzz around its service offering, with coverage from major media. The company believes in democratizing usage of TDA across industries. Certainly, its offerings are vast today and it has been able to show real performance, e.g. decreasing costs of providing healthcare. I will be curious to see how the technology will continue to grow and become more mainstream, and how it will co-exist with more traditional analysis tools.
1) Co-founder and CEO Gurjeet Singh, cited in Fortune article on Ayasdi
3) Lesnick, Studying the Shape of Data Using Topology, Institute for Advanced Study, Summer 2013.