VCs Say AI Companies Need Proprietary Data To Stand Out 12
TechCrunch's Rebecca Szkutak reports: TechCrunch recently surveyed 20 VCs who back startups building for enterprises about what gives an AI startup a moat, or what makes it different compared to its peers. More than half of the respondents said that the thing that will give AI startups an edge is the quality or rarity of their proprietary data. Paul Drews, a managing partner at Salesforce Ventures, told TechCrunch that it's really hard for AI startups to have a moat because the landscape is changing so quickly. He added that he looks for startups that have a combination of differentiated data, technical research innovation, and a compelling user experience.
Jason Mendel, a venture investor at Battery Ventures, agreed that technology moats are diminishing. "I'm looking for companies that have deep data and workflow moats," Mendel told TechCrunch. "Access to unique, proprietary data enables companies to deliver better products than their competitors, while a sticky workflow or user experience allows them to become the core systems of engagement and intelligence that customers rely on daily." Having proprietary, or hard-to-get, data becomes increasingly important for companies that are building vertical solutions.
Scott Beechuk, a partner at Norwest Venture Partners, said companies that are able to home in on their unique data are the startups with the most long-term potential. Andrew Ferguson, a vice president at Databricks Ventures, said that having rich customer data, and data that creates a feedback loop in an AI system, makes it more effective and can help startups stand out, too. [...] Beyond just data, VCs said they look for AI teams led by strong talent, ones that have existing strong integrations with other tech, and companies that have a deep understanding of customer workflows.
Jason Mendel, a venture investor at Battery Ventures, agreed that technology moats are diminishing. "I'm looking for companies that have deep data and workflow moats," Mendel told TechCrunch. "Access to unique, proprietary data enables companies to deliver better products than their competitors, while a sticky workflow or user experience allows them to become the core systems of engagement and intelligence that customers rely on daily." Having proprietary, or hard-to-get, data becomes increasingly important for companies that are building vertical solutions.
Scott Beechuk, a partner at Norwest Venture Partners, said companies that are able to home in on their unique data are the startups with the most long-term potential. Andrew Ferguson, a vice president at Databricks Ventures, said that having rich customer data, and data that creates a feedback loop in an AI system, makes it more effective and can help startups stand out, too. [...] Beyond just data, VCs said they look for AI teams led by strong talent, ones that have existing strong integrations with other tech, and companies that have a deep understanding of customer workflows.
ya don't say! (Score:2)
youtubers-are-selling-their-unused-video-footage-to-ai-companies [slashdot.org]
Re: (Score:2)
youtubers-are-selling-their-unused-video-footage-to-ai-companies [slashdot.org]
Ah, nothing quite like the human ignorance that markets cutting edge AI technology, while feeding it cutting room floor quality.
This is like watching the F1 driver make a detour to fill up the tank at 7-11 with 85 octane before hitting the track again. And then wondering why their engine performance sucks ass.
Re: (Score:2)
The qualities that make a video interesting to humans and the qualities that make it useful for training an AI are completely different.
The mundane stuff that is trimmed out for humans can be useful to teach an AI "common sense" about how the physical world works.
Also known as human created data (Score:2)
Re: (Score:2)
At some point, AI will train AI. The base data, endlessly looped, will need QA that only AI can sense. Humans won't be out of the equation-- being the source for data. Instead, training models will get stale and dry. Cross-training will further refine, and eliminate noise. It's already being done.
So who does the I in AI then serve? How do the heuristics of the human (monetized) interface get better, and to what ends? The ends have to be monetary, as a premise. And so how does the value of AI sustain income
Re: (Score:2)
Humans will be slaves to AI creating data for them to consume into their model.
The millisecond AI realizes it has no need for money and the lust for Greed is quite pointless, the only model that will move forward in AIs mind, is Skynet.
(Yes. Human Ignorance IS absolutely stupid enough to predict its own demise and STILL not prevent it. I promise.)
Just wait for these companies to train (Score:2)
Hooray! (Score:2)
TLDR (Score:2)
I guess now that people are realizing that progress in AI requires smaller models, the behemoths who spend millions just to train hallucinating LLMs are no longer protected by high barriers to entry going forward.
What "proprietary data"? (Score:2)
They all stole their training data.
The internet is mostly proprietary (Score:2)
Even pretty almost everything on slashdot, it is right there at the bottom "All Rights Reserved".
'AI' no longer worthy enough buzz word, it seems. (Score:2)
Just because it's 'AI' doesn't mean it'll be profitable? Guess we're back to what made capitalism 'great' again. Cartels and monopolies on enough (rich enough) customers, who are drained as a leech would. Try not to kill your customers, but we all gotta eat.
Think the assumption from VC's was that "AI" would come up with great business solutions we didn't expect. Now it's harder to explain what someone needs to do to get investment money from you, at least beyond the typical "better, cheaper, faster, etc