Apr 13, 2016 · 31m · 20vc
20VC: Why Machine Intelligence Will Eat The World Of Software with Roy Bahat, Head of Bloomberg Beta
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The 20 Minute VC, host Harry Stebbings interviews Roy Bahat, Head of Bloomberg Beta, exploring how machine intelligence is revolutionizing software, startup strategies against data incumbents, unique venture capital decision-making models, and the long-term impact of automation on the future of human work.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 24.3% of the talking time here. How this is scored →
Harry as informed peer 2.6 Guest teaching 3.6 Guest disagreement 2.1 Harry pushing back 1.4
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 18:22 Refusal to predict platform winners
When Harry asks Roy to predict the winning million-dollar use case on Slack, Roy flatly declines the premise, stating he has no idea and does not want to guess.
Hardest push from Harry ▶ 10:14 Challenging startup viability against data incumbentsHarry directly presses Roy on whether tech giants like Google and Facebook hold an unassailable data moat over early-stage startups.
Biggest teaching moment ▶ 7:21 Reframing AI as machine intelligenceRoy corrects Harry's broad pop-culture framing of AI, re-centering the conversation on machine intelligence as standard computer judgment applied to data.
Harry holds his own ▶ 10:14 Highlighting the data moat dynamicHarry demonstrates domain understanding by identifying how incumbent data access poses a critical structural barrier for emerging AI startups.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Roy Bahat's VC Career and Bloomberg Beta's Investment Model | 1 | 2 | 1 | 1 | Harry opens with standard background questions and banter. Roy shares how he accidentally entered venture capital and explains Bloomberg Beta's unique single-partner approval model. | |
| Machine Intelligence vs. Traditional Software | 2 | 5 | 3 | 1 | Harry brings up a teenage party analogy for AI excitement. Roy reframes the terminology away from pop-culture AI towards practical machine intelligence. | |
| Data Access, Incumbency Advantages, and Startup Strategies | 4 | 3 | 1 | 1 | Harry asks how startups can compete against tech incumbents with huge data advantages. Roy agrees data is the bottleneck and uses Textio to show how startups create virtuous data loops. | |
| The Impact of Open Source on Machine Intelligence | 3 | 3 | 2 | 2 | Harry inquires about open source, acqui-hire trends, and valuation spikes. Roy explains why open source is a powerful work model and notes that method-only AI startups face acquisition pressure. | |
| Platform Ecosystems and Emerging Startup Frontiers | 2 | 4 | 4 | 2 | Harry presses Roy to predict the first million-dollar Slack platform use case. Roy explicitly declines to speculate, arguing that first rarely equals best in new platform ecosystems. | |
| The Future of Work, Automation, and the 'Human Corner' | 4 | 4 | 1 | 2 | Harry cites stats on white-collar job automation. Roy discusses economic safety nets like UBI and introduces his human corner theory regarding future labor. | |
| Honesty Box Quickfire Round | 2 | 4 | 3 | 1 | During a quickfire round, Roy critiques traditional early-stage VC mindsets while detailing how Bloomberg Beta supports founders through mistakes. |
Statements from this episode (18)
Bloomberg Beta invests if a single partner says yes, ignoring consensus
“Any one person says yes on our team we do a deal.”
The best startup ideas are highly polarizing, not consensus-driven
“The best ideas are highly polarizing, and so we've set up a model to try to harness their polarizing effect to make the best decision, which is to say we want to say yes when there is one person on our team who is very excited about it.”
Bloomberg Beta avoids the term AI due to sci-fi humanoid connotations
“So we actually try to avoid talking about AI. And the reason is AI has come to mean both the cluster of all those techniques and the Like replicating human intelligence with something that feels humanoid and, you know, is out to get you like in Ex Machina or s…”
Machine intelligence will eat software while software eats the world
“Yes, I think machine intelligence will eat software. While software is eating the world.”
Big data produced valuable technologies but very few valuable standalone businesses
“The, you know, the quote-unquote big data thing Turns out to have produced a number of very valuable technologies, not that many valuable businesses and part of it is because the data on its own, as everybody knows, is, you know, it's just information.”
Proprietary data access is the primary bottleneck in machine intelligence development
“At least today, access to the data is the limiting factor. Which means, by the way, not only do companies like Facebook and Google have an advantage, but pretty much every old industrial incumbent, if it has a way of getting at its data and making use of it, a…”
Horizontal black-box AI startups face severe go-to-market challenges without consulting
“The warning is, if you're startup is, hey, I've got a black box that does machine intelligence. Let me try to apply it to a bunch of different situations. Then I think the going is very hard. You know, you probably have to start with some consulting services. …”
Open source is the most important work collaboration method ever invented
“Open source is probably the single most important new method that we as a working culture have invented, period.”
Bloomberg Beta published its fund operating manual publicly on GitHub
“We published our manual to GitHub and released it under an open source license”
Method-only AI startups without specific applications usually end up as acqui-hires
“If what you're talking about is, I guess I'd call them method-only companies, you know, some technology that you think can be applied to a wide variety of situations, and you're not really building around any one of them, then yes, then I do think that is aqua…”
First-mover companies consistently lose to those who build the best product
“I mean, I think we tend to be obsessed with first in the technology industry, and first loses to best every time. You know, the first iOS app companies were not the best iOS app companies, you know, platform after platform, and I think that's exactly what we'r…”
Roy Bahat identifies eldercare as the number one untapped startup opportunity
“We don't invest in the space of things that serve people's personal lives, but if I did, I think services for the elderly are the number one untapped opportunity, and just wait till we have more sixty-five-year-old tech founders, because they'll see the opport…”
Society must reconsider work and subsistence within 10 to 20 years
“And my view is, A, that we probably will need to reconsider the relationship between work and subsistence at some point in the next 10 to 20 years.”
Future human employment will be defined by demand for handmade goods
“What jobs will human beings do? Well, they'll do the jobs when other people want to buy something specifically because it was made by a person. And we have a word for this, which is handmade.”
Roy Bahat argues the typical venture capitalist is disastrous for startups
“I think there are many wonderful VCs. I just think it's that the typical VC is disastrous.”
Bloomberg Beta backs founders who reject advice to step down
“If we thought that there was a better alternative than the founder for the business, we'd tell the person that. And then if they said, no, no, no, thanks. That's nice. That's your opinion, but I'm going to keep going. Generally, we would back them to keep goin…”
Bloomberg Beta encourages founders to shop term sheets without exploding deadlines
“When I give people a term sheet, I say, go shop it. It's not going to explode. I only want you to work with us because we're the best choice for you, not because I've put you under some time pressure.”
Early-stage investing requires finding one reason a startup could be an outlier
“What we look for is, there are a bunch of things that are tick boxes. You know, do we trust the founders? Are the deal terms fair? You know, is it in scope for us? But then really what you're looking for is, one reason to believe the startup could potentially …”