This is an age of wonder.
The changes driven by the mix of SmartPhone (universal access), Cloud computing (affordable, on-demand, computing), are well underway, and affecting every industry. Even with the changes seen so far, and the amazing number of companies that have emerged to drive this change, is it still only the beginning. More and more of our assets are able to benefit from ongoing exponential improvements in speed, capability and more. (For more, check out the Singularity Institute.)
So how do we use all this new computing capability?
In part, by finding practical applications for technologies pioneered in the field of ‘artificial intelligence’, solving more ‘creative’ problems. As I’ve noted previously, in general, it’s important to think of these technologies as providing Augmented Intelligence, making us better, rather than get caught up in the epistemological trap of fear of the AI. In any case, we are seeing a massive rise in interest by investors in companies that make ‘smarter’ computing around Machine Learning, Natural Language Processing, and Vision.
In most cases, these technologies will be used extremely pragmatically. In ways that just make sense, that simplify our days, or indeed, that simplify error prone processes. That brings me to Coseer.
Coseer is a company that joined the Citrix Startup Accelerator – Innovators Program last year.
They take exactly this pragmatic approach to using new technology to solve real problems. Here’s the Coseer CEO:
At Coseer we believe humans should focus on creativity and innovation, while technology handles everything else. Our technology automates and scales tedious language-related tasks in enterprise workflows. These include extracting summaries, automatically filling up databases, answering natural language questions, or others that currently require human processing. Our modular approach lets us help Fortune 500 enterprises as well as startups for a variety of problems.
Coseer’s technology consists of multiple modules that emulate simple human tasks, like extracting all key-value pairs from a text e.g. “Coseer’s age is 2 years.”, extracting relationship statements e.g. “Coseer – is based in – San Francisco.”, etc. Once emulated, these tasks can be automated e.g. automatic checking a website for new information; scaled up, e.g. reading through millions of documents, and fed into advanced AI analytics. Examples of these analytics include deduplication, anaphora resolution, clustering, correlation with other datasets, and training some machine learning algorithm. These modules and analytics can now be configured into a very wide range of solutions.
For example, our Coseer Express product for Investment Management starts with extracting all the key messages from millions of articles. These messages are normalized using anaphora resolution, e.g. “The company sells this product in the country.” will be resolved to “Coseer sells Coseer Express in the US.” to make it objective. The objective messages then run through a deduplicator and correlated with various known sets specific to Investment Management to bubble up the most important sentences. These messages are then placed in a taxonomy and reported to the user.
Using similar constructs, Coseer is being useful in many other industries and functions. For instance in eCommerce, Coseer summarizes all the reviews so that key phrases can be reported on a mobile screen or via tweets. For Marketing, Coseer predicts the odds of success of any promotion. For Insurance, Coseer automates the routine Q&A via emails, tweets and SMSs, and escalates to humans only when necessary. For Sales, Coseer tracks all developments for every customer and executive, bringing equal focus to all accounts, big or small.
Technologies like Coseer are a very pragmatic start to what’s expected to be a flood of interesting new capabilities that will continue the transformation of our industries wrought by cloud and mobile. At least, that’s what I see. What do you think?
Dr Michael Harries, Chief Technologist, Senior Director, Citrix Startup Accelerator