Welcome to AdExchanger Talks, a podcast focused on data-driven marketing. Subscribe here.
Before Jeremy Fain co-founded Cognitiv in 2015, he worked a long stint at Rubicon Project, where he noticed something off about the bidding algorithms used by different demand-side platforms.
“We saw the standard bidding algorithms all over the place,” he says. “That didn’t jibe with me.”
After all, Target customers behave and react differently than Walmart customers. But even when ad tech companies like AppNexus began letting clients install their own custom algorithms, there were still problems.
Algorithms require tremendous expertise and manpower to build and scale: “You still have to guess what variables to use.”
In this episode, Fain talks about how deep learning can automate algorithm building. He discusses how Cognitiv’s focus on big data and deep learning – and departure from more common artificial intelligence disciplines like natural language processing and computer vision – let it carve out a niche.
That expertise made Cognitiv a key partner to IBM, its exclusive US sales partner, in that it powers numerous applications in Watson Advertising.
So maybe the next step is an IBM acquisition?
“That would be a wonderful exit, of course,” Fain says.
Also in this episode: The biggest challenge starting an ad tech company in 2018. Is it a problem that AI is no longer the talk of the town? And what’s the state of deep learning talent in Maryland?
This post was syndicated from Ad Exchanger.
More Stories
The IAB Predicts Social Video Will Overtake CTV This Year
Best Buy and CNET Are Combining Their Ad Inventory
Buyers Are Wasting Money on Alt IDs While Cookies Still Persist