The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 31 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Wang: Scale AI will hire purely on merit without demographic quotas
“Yeah, so MEI, we basically rolled out this idea of merit, excellence, and intelligence and the basic idea is in every role we're gonna hire the best possible person regardless of their demographics and we're not going to do any sort of you know quota-based opt…”
Wang: Pure AI model renting will be a mediocre long-term business
“Lack of pricing power, let's say, on the pure model layer certainly indicates that renting models out on their own may or may not be the best long-term business. I think it's likely to be a relatively mediocre long-term business.”
Ben Scharfstein estimates enterprise AI solutions are 70 percent unglamorous integration work
“What does the end-to-end solution look like? Which only 30% of it might involve AI, and the rest is that schlep that you just have to do to have an impact.”
Wang: Far fewer enterprise AI proofs-of-concept reach production than expected
“Much, much fewer of the POCs have made it to production than I think than I think the industry overall expected, and I think a lot of enterprises are looking at it now, and, you know, the doomsday that they thought might have happened hasn't really happened.”
Wang: Founder CEOs must remain embedded in critical decision-making loops
“The reason that you are a good founder CEO is because you make very good decisions over and over and over again over an extended period of time. And to pull yourself out of those decision making loops is, you know, would be kind of crazy.”
Wang: Letting new external executives build large teams quickly causes ruin
“I think this almost always results in ruin. I think that this isn't to say that you can't hire executives from the outside, but I think what you need to do when you hire executives from the outside is you really like, you like, they really get steeped in how t…”
Wang: AI labs will diverge as LLM progress shifts to research
“I think we're entering a phase where the research is going to start mattering a lot more. Like, I think there will be a lot more divergence between a lot of the labs in terms of what research directions they choose to explore and which ones ultimately have bre…”
Wang: Current AI agents fail because the internet lacks agent training data
“Agents has been the buzzword for the past two years and basically no agent really works. Well, you know, it turns out there's just no agent data on the internet.”
Scharfstein: Forward-deployed engineering models are not scalable outside enterprise sales
“If you're not selling to enterprises, it's difficult to think about this as a scalable business model.”
Scharfstein: Solving end-to-end enterprise problems unlocks 10x larger contracts
“Like your job is not to build a product. Your job is to solve a problem. And that is what enterprises are going to expect. That is what's going to unlock 10 X bigger contracts with them.”
Scharfstein: Difficult 10x engineers belong on platform teams, not client teams
“That's like very hard to work with, but they're a 10 X engineer. Those people should be on your platform team and not on your forward deploy team.”
Wang: Next AI phase defined by data production as public data depletes
“We're kind of hitting this wall where we've leveraged all the publicly available data. And so one of the hallmarks of this next phase is actually going to be data production.”
Wang: Scale AI grew revenue 6x while keeping headcount flat
“So over the past few years we've Basically kept our headcount flat. I mean, we've grown it very slightly as the business grown, but the business itself is, you know, five X, well, six X, like, you know, the business has grown dramatically.”
Scharfstein: AI aims to augment human work, not replace software
“What we're moving in AI is that we're not trying to replace software. We're trying to augment or automate human work.”
Scharfstein: Enterprise AI companies must become systems of intelligence
“I think that's the key thing is you want to become a system of record system of work and eventually become a system of intelligence where you actually do the work.”
Scharfstein: Forward-deployed engineering roles are a factory for founders
“I think it's like a factory for founders, whether or not it's in the product role or it's in the Ford deployed engineering role or AI role, because these are people that are just doing customer discovery while they're building, which is like what being a found…”
Scharfstein: Spending 3 days on-site accelerates enterprise AI delivery by 3 weeks
“Sitting with the customer, you know, flying to six hours away and saying, I'm going to spend three days with you is invaluable. Speeds you up three weeks.”
Ben Scharfstein says enterprise AI lags public state-of-the-art by 18 months
“And I would say it was probably like an 18 month lag between what you see on Twitter and what's the state of the art and what is actually working in enterprises.”
Scharfstein: Rapid AI shifts make custom enterprise builds superior to static products
“That's because the industry is changing a lot. Every day, ah, it changes, you know, GPT-V comes out, and it's different than, you know, it was three months ago, and so what we need to actually build in product changes, and so we've just said, you know what, th…”
Scharfstein: Scale AI allows large enterprise clients to retain their IP
“Oftentimes the big customers want to retain the IP. We're very happy to allow them to retain that IP because, you know, it's core to their business.”
Scharfstein: Forward-deployed engineering must build durable software, not consulting
“And the key is that it should be the customization and a wedge into installing your software that is durable over time. It's not worth just doing this as, you know, there may be great consulting businesses to build, but that's not, you know, the pot of gold.”
Scharfstein: Enterprise AI shift demands forward-deployed engineering for custom software
“And what we're seeing now with this platform shift is enterprises still need all of that customization. They need all of the feature set, but it doesn't exist in the software and doesn't exist anywhere. And so that's the value of this forward deployed motion i…”
Scharfstein: Forward-deployed engineering must feed custom client fixes back into platform core
“Really the mandate of the forward deployed engineer is to do that. You do need to say, okay, we have 60% out of the box, but that last 40%, maybe it's a data integration, it's a visualization, it's an agent that we haven't built yet. You do need to do that wor…”
Scharfstein: Startups must retain IP and standardize features when serving small clients
“I think the smaller the company and the smaller the customer, the more that you need to retain all of the IP and build it back into the platform.”
Scale AI's Ben Scharfstein: AI startups must avoid vanity revenue metrics when delivering services
“One is I think you just have to be very honest and very sober about your revenue when you do that. And you say, Hey, like we were able to chart, get to ten million dollars in revenue, but it's, Consulting-ish revenue. Our repeatable aspect is two million, and …”
Scale AI employs over 100,000 global contributors for AI data tasks
“We have contributors, you know, a 100,000 plus contributors around the world That do these tasks.”
Wang: Past three years of LLM progress driven by execution, not research
“For the past two-ish years or the past maybe three, four three years, let's say three years it's almost been more about execution than anything. It's a lot of just engineering, like how do you actually have large scale training work well. How do you make sure …”
Sherman Wu: Early Quora team included future Scale AI and Perplexity founders
“A bunch of the perplexity team was there. Dennis, Dennis was on the feed team with me. Johnny Ho Jerry Ma. And then Alexander, the scale, you know, like was there. He was there between high school and college.”
Scharfstein: Scale AI's enterprise business found PMF in past 12-18 months
“There's been an enterprise team at scale for a long time, but really just in the past 12 to 18 months has really taken off as something that's found kind of product market fit.”
Scale AI's enterprise applications make up half of its total business
“The other half of our business is the application business”
Scharfstein: Scale AI splits forward-deployed teams into software, ML, and product
“At scale, we break down the forward deployed role, not just a forward deployed engineers, which, you know, Palantir made famous, but we also have forward deployed product. That's the team that I lead and forward deployed machine learning engineers or an applie…”