When Bloomberg first built the terminal system, back in the early
1980s, most of its customers — mainly finance professionals —
didn’t have computers on their desks.
The internet was not yet a commonly-accepted technical protocol
for networking and hardware of its kind hadn’t been seen before.
So Bloomberg’s engineers had to go about inventing the tech
themselves — from the set of instructions to carry data across a
network, to custom-built hardware so traders could use a
keyboard, and monitors you could stack.
It created a great culture of invention at Bloomberg, which has
more software engineers than journalists. But cultivating that
culture came at a small cost.
“A number of executives at Bloomberg realized technology was
being developed in the startup world they were only seeing once
it was mature enough to be ready for Bloomberg, but that was
often too late to be able to fully understand it. So they just
wanted greater awareness of what startups generally were doing,”
says Roy Bahat, head of Bloomberg Beta.
Bloomberg Beta is the company’s in-house early-stage fund, which
has $150 million under management. The firm, which only makes
investments at the Series A round or earlier, focuses on
companies involved in changing the way people will work in the
Bloomberg approached Bahat — a former News Corp vice president,
IGN entertainment president, and cofounder and chairman of gaming
console OUYA — to head up the fund in late 2012.
“At first my reaction was: This would definitely be a horrible
idea for Bloomberg because most of my experience with corporate
investors was … problematic to say the least,” Bahat said.
But Bahat was won over by Bloomberg Beta’s principles that it
would only invest in companies to make money — rather than trying
to back companies with the intention of Bloomberg working with
them, which is where problems can arise. The firm claims it has a
Net Promoter Score — a measure of satisfaction between -100
and +100 from its portfolio founders — of 82.
“We joke that if you add ‘.ai’ do your domain name, you get a 20%
bump in valuation”
Within its future of work focus, Bloomberg Beta has honed in on a
particular area: machine intelligence/AI.
Bahat admits he “completely missed” the trend in the beginning.
“One of my partners came to me and she had done some analysis of
big data infrastructure, and she said: ‘Now that companies have
amassed all this data, the next problem is: What are they going
to do with all of it?’ She said what they are going to do is AI,”
Bahat said. “I kind of thought she was taking a page out of a
science fiction novel. I remember saying something to the effect
of: ‘Well that sounds wacky, but if you want to spend time on it,
by all means’.”
Bank of America Merrill Lynch predicted last year artificial
intelligence analytics will represent a $70 billion
market by 2020. And with big-name companies like IBM, Google,
Microsoft, and Facebook making
big bets on AI, everyone wants a piece of it.
AI has become so commonplace it’s hard to find a company that
doesn’t use some form of machine learning technology in some way,
somewhere, Bahat said. But the rise of AI, particularly over the
last three to six months, does mean many more companies are
trying to pass themselves off as an AI business, when in fact
they’re something else.
“Certainly a lot of companies market themselves as AI. We joke
that if you add ‘.ai’ do your domain name, you get a 20% bump in
valuation,” Bahat said. “In the same way that five to seven years
ago, all these companies called themselves ‘big data’ companies.
These things are fashion and we were fortunate to get ahead of
the fashion. The fashion will pass us. I hope to remain working
on it long after it’s fashionable.”
Bahat has big hopes for AI.
“AI … surely will be a trend at least on the size of big data.
It almost certainly will be a trend on the size of mobile. It
might be a trend on the size of the internet. And maybe, just
maybe, it’ll be a trend on the size of software; that the
software before machine intelligence and after will be two worlds
that are very different from each other,” Bahat said.
What to look for in an AI startup
Bahat says he created his own metric when it comes to AI
conversations: Time To Her (TTH), meaning how long will it take
before the movie “Her” comes up. As far as business metrics
are concerned when trying to assess which AI startups to invest
in, he says this is a little trickier.
“It’s so hard because venture is kind of like a sporting game,
where you start playing and you only learn the final score nine
to 11 years later — and the score at the beginning of the game
bears relatively little correlation to the score at the end of
the game,” Bahat said.
Bloomberg Beta tends to ask two questions of its potential
investments: 1) Do you have access to your own user? 2) Do you
have access to a data set that’s yours? If they have both those
things, then they can create a virtuous cycle where the user
contributes the data, the data gets better, and it makes for a
better user experience.
Startups that have fit the bill and have recently earned
investment from Bloomberg Beta include: Textio, a platform
that lets recruiters upload a job description and uses machine
intelligence to make recommendations about improvement to the
wording in order to attract better, more diverse applicants;
AppZen, an app that uses AI to detect peculiarities in employee
expense accounts; and Orbital Insight, a company that buys
satellite data to estimate the size of markets — for example,
measuring the growth rate of the construction industry in China
by inspecting the shadows on buildings.
Bahat said he’s also biased to taking meetings with founders from
under-represented communities because he thinks there are more
chances of spotting a winner. The founder of Orbital Insight, for
example, is in his 50s.
For the past three years, Bloomberg Beta has partnered with
Berkeley’s Haas School of Business to predict the 350
“future founders” in the Bay Area and New York who will go on
to start venture-backed companies. Last year, eight of the
people the study predicted would become founders went on to start
their own companies. Bloomberg Beta and Berkeley believe their
predictions are now 50 times better than chance.
One of the surprises thrown up by the
most recent year’s study was the make-up of the average
entrepreneur: One in five of the predicted future founders were
women, they were more likely to “have enough education, not too
much,” and the group was “much older” than the usual startup
founder stereotype, according to Bahat.
“I think one of the most important dynamics will be older and
older people doing things we previously thought of as a young
person’s game,” Bahat said. “I’d like to see the [startup
accelerator] Y Combinator of people over 55 — maybe Y Combinator
will become the Y Combinator of people over 55.”