The Silicon Review
It’s maddening. Here we are in the age of Go-conquering robots and self-driving cars, yet up and down the world – and even in Silicon Valley – highly skilled workers are still wasting their time on tedious tasks. From the marketers compounding multi-platform analytics data, to the lawyers redacting treatise after treatise, to the product designers losing days sitting in on market research events, white collar professionals increasingly feel frustrated by workflows that don’t exercise their most valuable expertise.
The most recent statistics suggest less than half of us are satisfied by the stuff we do each day – and a similar proportion are too talented to be doing it. Coseer CEO, Harvard Business School alumni and former technology investor Praful Krishna used to a member of this army of overqualified and underutilized human capital.
“When I came up with the idea for Coseer, I was working at a hedge fund,” says Krishna. “Day after day I had to spend a crazy percentage of work hours trekking this endless paper trail, going through mountains of analyst’s reports, regulatory filings, news clips, social media. It wasn’t time wasted but it was time that could have been better spent.”
Krishna wanted to employ his acumen in decision-making rather than researching. If he could find an artificial intelligence that would automate all the legwork, he would be able to focus on the strategic tasks that ladder up to better investments.
“People – or the creativity and judgment for which our intellect is famed – are an organization’s most valuable resource. So why are we all stuck doing tedious jobs that don’t utilize the part of our brain that makes us valuable?”
The answer to that question is mess: most organizations are actually very unorganized. They require humans to sift through all the detritus they and the external world produce, and put it together in a productive way. Getting a computer to emulate this kind of information processing is a monumental task: “What I was really there to do was use my judgment to make buy and sell decisions. But I had to make sense of all these messy, unstructured datasets first and, at the time, no one could automate that kind of thing efficiently,” says Krishna.
He soon realized his idea went way beyond putting highly skilled people to better use. If a business owned one AI that could rapidly train and apply to any process in its organization, it would also own a phenomenal advantage over its competitors: it could make faster actions and decisions, optimize portfolios and react to market fluctuations 24/7, reduce labor costs, achieve greater scale, and uncover product and service innovations in the blink of an eye.
That was 2012. Fast-forward five years and Krishna’s company Coseer is one of Silicon Valley’s fastest growing startups, with clientele in almost every sector you can think of; consulting, consumer, finance, industrials, healthcare, legal, life sciences, publishing and technology. It no longer matters how messy an organization is: the AI he and his team developed emulates human thought process to such a degree it can automate an abundance of the complex-but-tedious workflows humans hate to do.
“We specialise in text-driven workflows. Our AI can collect, analyse and make decisions about any textual information you put in front if it.”
Krishna says this is helping P&L decision-makers transcend conventional performance trade-offs, such as having to choose between efficiency and quality. With an automating AI, companies can have their cake and eat it.
“They can turn up insights that would take the most intelligent human more than a lifetime of reading and research to discover. They can turn these insights up round the clock, at a fraction of the cost of a human resource, without any manual or fatigue-related errors, cognitive bias or risk.”
This is not just automation. It is intelligent automation; interpreting facts, taking decisions and adapting to change. For Krishna and his clients, reaching this stage of sophistication marks a monumental step forward: “Digital technology only knows ones and zeros. Getting it to capture something as abstract as human thought is unimaginably difficult.”
Coseer’s approach was to use a branch of artificial intelligence known as cognitive computing – the “cognitive” part is how it makes sense of inputs traditional computing approaches have a very hard time with. Utilizing a combination of natural language processing, real-time computing and machine learning algorithms, Coseer can decode the messy, ambiguous outside world, and structure the resulting information so more traditional computer programs can take action.
“Effectively, we are automating what can’t be automated,” Krishna says.
Of course, we’ve all heard this story before – so why aren’t we all relaxing while the robots do the hard work? Why are so many people still wasting so much time on repetitive mental labour?
Aside from cognitive computing technologies still being in their infancy, the ones that do work are so expensive, most businesses are priced out of the game. They take months or even years to train to an acceptable level of accuracy and require huge computing power. Training time for an IBM Watson application can be as high as three years and cost upwards of $5 million. Other AI companies like Digital Reasoning, IP Soft, and FlatIron offer more affordable and agile solutions, but their applications are so specific they cannot impact on an organisational level.
Coseer eclipses both of these blindspots by being a general AI that can rapidly apply to any part of an organisation, at a radically lowered price point.
Krishna says, “We were not satisfied with these cumbersome, single use case, legacy applications that saturate the B2B AI market. We wanted something that was truly transferrable. Something that, like Windows and ERP, could impact and transform across functions, industries and job roles.”
He calls his solution Tactical Cognitive Computing. It combines high accuracy (98%), low training times and an accessible structure to give businesses big and small an automating solution that doesn’t break the bank.
Coseer specializes in text, so if a business wants to automate information processing from other inputs, like videos or audio transcripts, they’ll have to look elsewhere. Krishna sees this as a benefit.
“Because we focus on one form of messy data, text, we can create a solution that is more accurate than anything else on the market, and provide it a much more affordable rate.”
Pressed to divulge the reasons behind his company’s rapid growth, Krishna says his company makes one clear promise to every client: “You don’t need to allocate years of planning and budget allocations to this – you can deploy a solution within four to 12 weeks - if its not going to work, you know it early. This is what we mean by ‘tactical’ – you can apply our AI exactly where you need it, when you need it.”
There’s still much more to accomplish. Due to demand from multinational business Coseer is currently developing an evolution of its AI that will be able to understand and converse with humans in languages other than English. And, perhaps most tantalizingly of all, Krishna and his team are working towards building a machine that really does think like a human – one that will be highly accurate without any need for training.
“People still think of this stuff as being science fiction,” says Krishna. “It’s not. It’s already transforming industries. If businesses don’t start thinking seriously about cognitive automation, they’re going to be left behind.”