If you’ve bought something on eBay, searched a home improvement store’s website, listened to music on Spotify, or watched a video on YouTube, you’ve come in contact with CrowdFlower.
“People don’t realise it, but you touch CrowdFlower every day,” says founder and chief data scientist Lukas Biewald with more than a little pride.
Today CrowdFlower employs more than 100 people, has “tens of thousands of users and thousands of customers,” including Autodesk, Google, Facebook, Twitter, Cisco, Github, Mozilla, VMware and others.
CrowdFlower has raised $US58 million and is valued at $US110 million by its investors, according to PitchBook, the database that tracks such things.
The future is looking bright. Biewald wouldn’t share a sales figure but said, “We almost tripled revenue this year and we work with almost all the big machine learning teams in Silicon Valley and financial services.”
Companies use CrowdFlower after they write a machine learning algorithm to automate a task.
Computers learn by example. Training them requires gargantuan numbers of examples. CrowdFlower helps companies get the training data they need for their algorithms from their own stores of data. It then helps the program figure out what tasks it can do itself and what tasks require human intervention. Companies use it to improve catalogue search results, approve photos, support customers, fight online bullying and so on.
A phone call from Travis Kalanick
Today machine learning and AI are the hottest technologies. But back when Biewald founded the company in 2008 at age 25 — the height of the financial crises — machine learning was considered a boring, geeky thing, pooh-poohed by investors.
“Machine learning was really not cool,” he laughs now. When investors heard the term machine learning, “they thought, ‘research project,'” says Biewald, who got the idea for the startup from his time working at Yahoo.
The other thing that wasn’t particularly prized back in 2008: being a startup founder. This was a couple of years before the launch of 500 Startups, and a time when Ycombinator was a fledgling thing.
“It wasn’t cool to be a young founder. People were saying that Zuckerberg was going to get fired. I remember my parents thinking, ‘Do you actually have a job?,'” he says.
And while Biewald clearly knew the technical problem he was trying to solve — supplying training data for AI apps — he had no clue on how to run a company.
He was searching for customers with nothing more than a website with his phone number.
Then one day the phone rang. It was a guy named Travis Kalanick, who had seen the website and thought the idea was amazing.
“We met at the food court at the mall. I had no idea who he was,” remembers Biewald.
That was for good reason. Kalanick wasn’t anybody back then. His track record was Scour, a file sharing startup that went bankrupt in the face of an insane $US250 billion lawsuit from movie and music industry trade groups, and Red Swoosh, another file sharing company that was eventually sold to Akamai for $US19 million, but after a lot of struggle.
Today, Kalanick is best known as the cofounder and former CEO of Uber, the ride hailing giant that has grown into the world’s most valuable tech startup with a $US69 billion private market valuation.
Before Uber took off though, Kalanick was angel investing and running what he called The JamPad, the name for his San Francisco home. He opened his house to entrepreneurs to hang out, eat healthy food, sometimes sleep on the couch and learn about running their companies from other invited guests.
After that first food court meeting, Biewald and Kalanick became friends and Kalanick began to mentor Biewald.
Kalanick spent hours helping Biewald perfect his investor pitch. And he was loaded with practical, tactical advice like which words to use for each investor and which ones not to use (for instance, he told Biewald to take the “machine learning” out of the deck because investors “wouldn’t understand it.”). He was full of other tips like how to get into important trade conferences for free (find a way to volunteer) and how to attend other shows on a shoestring budget.
More than that, Kalanick taught Biewald to operate his startup with a sense of urgency. He advised him to set a deadline for his first round of funding and tell investors to be in or out by that date. When investors or customers seemed interested, Kalanick told him to follow up immediately, not even wait an hour, preferably in person.
“I didn’t have a car. He would drive me to investor pitches,” Biewald remembers.
Kalanick himself invested about $US50,000 of seed money, one of the first to invest, Biewald said.
By the end of 2009, Biewald and his startup were on their way. He had raised a $US1.15 million seed round and had landed some well-known angels like Dave McClure, early Facebook employee Jeff Hammerbacher (who went on to co-found Cloudera), and FF Angel (operated by Peter Thiel’s Founders Fund).
“There were other people that helped me,” he said. “But the thing that was unique about Travis’s help is that it was for long periods of time. Like every day, every week. That’s key. It’s easy to drop in and give advice. But like, when nobody cares and you’re not making money, that’s the challenge. He really cared.”