# Serving varied factories forces robots to generalize

Markdown copy of https://realalpha.app/s/serving-varied-factories-forces-robots-to-generalize/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Josh Gruenstein](https://realalpha.app/people/josh-gruenstein/) · Josh Gruenstein is cofounder and CEO of Tutor Intelligence.
- Host: [Theo Jaffee](https://realalpha.app/people/theo-jaffee/) · Co-founder & host, MTS · One of the first members of a16z's new media team; earlier hosted the Emergent Ventures-backed Theo Jaffee Podcast.
- Host: [Sophia Dew](https://realalpha.app/people/sophia-dew/) · Host, MTS
- Show: MTS
- Episode: [Robots Will Proliferate Daily Life in 5 Years | Josh Gruenstein](https://realalpha.app/v/xNKXt3Hb8Tk/), Nov 3, 2026
- Moment: 15:22 to 18:12 of the episode (3 min), playing on https://realalpha.app/s/serving-varied-factories-forces-robots-to-generalize/; the episode from there: https://realalpha.app/v/xNKXt3Hb8Tk/?t=922

## Quote

> “the only way you can do all of that work is with a **truly general purpose robot**.”
> — [Josh Gruenstein](https://realalpha.app/people/josh-gruenstein/) at [16:54](https://realalpha.app/v/xNKXt3Hb8Tk/?t=1014) · MTS

## Takeaways

1. Deployment data depends on manufacturing capacity and winning customers.
2. Serving very different factories tests a robot’s generality.
3. Custom engineering on every deployment prevents scale.

## About this segment

Josh Gruenstein says robotics advances in a loop of deploying robots, collecting data, building capacity and winning customers. He argues that serving very different factories forces robots to generalize without custom engineering.

From “Robots Will Proliferate Daily Life in 5 Years | Josh Gruenstein”, Nov 3, 2026. [Watch the full episode](https://realalpha.app/v/xNKXt3Hb8Tk/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[15:26](https://realalpha.app/v/xNKXt3Hb8Tk/?t=926) Would you say data is the biggest bottleneck at getting robots manufactured and diffused into these factories at scale, or is it something else? I think it's all a loop. You deploy a robot, and it generates the data that you need to train models. It's like, okay, what's bottlenecking you from doing that? Well, you had to manufacture that robot or you had to get that customer. And what blocked you from getting that customer?

[15:47](https://realalpha.app/v/xNKXt3Hb8Tk/?t=947) It's like, well, maybe you didn't have the manufacturing capacity or you couldn't train the model. So it's all this circle of life, right? And if you can get velocity anywhere in that flywheel, you can build. data, the supply chain, and the distribution, that really pushes the needle in robotics. Yeah. One thing you mentioned is you guys talk to a lot of customers, which you would be shocked how rare that is. A lot of people are building intelligence, like, oh, this is going to solve X, Y, and Z, and yet you've never talked to anyone working on X, Y, and Z about how it actually gets into their company. One thing, though, is when it comes to factories, when it comes to manufacturing, every single industry, every single company, every single factory is completely different from the other. And so are you guys working with like one specific type of industry

[16:34](https://realalpha.app/v/xNKXt3Hb8Tk/?t=994) or just everyone across the board? No, we work with everyone. And you're right, that's very hard. It kind of sucks, right? But that's how you know you're building a general purpose robot. It's this beautiful forcing function because the only way to serve the crayon manufacturer and the peanut butter manufacturer and the crayon manufacturer or the fork manufacturer and this warehouse and that warehouse, the only way you can do all of that work is with a truly general purpose robot. If you're not building a general purpose robot, then you need engineers on every deployment, you need to solution custom software and custom hardware, and you just can't scale. It doesn't work. We deploy to a really wide diversity of businesses, and we feel like that's critical to build the technology and the product we actually

[17:19](https://realalpha.app/v/xNKXt3Hb8Tk/?t=1039) want to build. Yeah. I feel like there's these two debates around, should people be building new factories from scratch that automatically have robotics already there? Or should you augment existing factories to have more autonomy within them? And then they keep a lot of their existing machines, but they're just more improved with all the new technology happening. What are you seeing as the more efficient way? Both. Do both. More factories, more better. Absolutely. I think you can make the analogy in software, right? We had all of these traditional businesses that predated the internet and the cloud, and they went through this digital transformation. A lot of those are still very successful businesses. You also had whole new categories of businesses transforming markets that couldn't have possibly existed before the cloud. I think we'll see the same

[18:01](https://realalpha.app/v/xNKXt3Hb8Tk/?t=1081) thing with robotics. Interesting. All right. So today was a big day. I know you say, me wasn't as big of a day where you're like, this is just the capstone to this big announcement.