Real Alpha
Watch modeSubscribed

Training Data · 2 min · Oct 20, 2026

Robot data factories need adaptive quality gates

JG

Josh Gao

Founder & CEO, Mecka AI

  • Sonya HuangHost · Partner, Sequoia Capital

    Works with OpenAI and Anthropic; co-wrote Sequoia's essay on generative AI.

About this segment

Gao says quality standards shift as new real-world edge cases appear, so Mecka adapts its checks and escalation process. He cites contributors submitting copied McDonald’s POV videos, then stresses on-site oversight and incentives.

From “Inside the Robot Data Factory: Mecka's Josh Gao on Sensorizing Humans to Train Robots”, Oct 20, 2026. Watch the full episode

“at one point we have folks recording YouTube videos like POV working at McDonald's and trying to submit that as contribution pretty much.”
JGJosh Gaoat 17:07 · Training Data17:07

Takeaways

  1. Quality standards change as new edge cases appear.
  2. Fraud can include relabeling public POV videos as contributions.
  3. On-site oversight and incentives support data quality.
Watch 2:25 from 15:19

In this episode

Full episode
  1. 31:56Robot training is converging on a mixed data recipe3:56
  2. 12:07Running a robot data factory requires custom hardware and logistics2:45
  3. 19:14Robotics can unlock commercial tasks before reaching generality2:40
  4. 8:29Egocentric data gave Mecka a scalable starting point3:27

  5. 15:19Robot data factories need adaptive quality gates2:25

  6. 21:57Constrained jobs will come before household robots2:50Up next
  7. 24:38Robot integrators can improve installed bots with data3:31

Transcript

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

  1. 15:25

    No, I'm kidding. Please do. Okay. Interesting. Like I've heard that, you know, there's a lot of quality issues in robot data collection generally, and that a lot of the hours of robot data sample that get sent in end up not being accepted. Why is that? And what are you doing about that problem? This lesson came from, even originally, the different types of robot data. There have been these very large-scale robot factories that have spun up from just large-scale warehouses repetitively doing tasks, trying to scale up teleoperation data.

  2. 16:00

    Some of the learnings there came from the research in that there is some threshold of diversity where you oversaturate how much robot data you need or how much data in general you need. And across human data, quality, you can imagine, is actually quite a bit of a moving target as we grow. The amount of edge cases that you have in the real world that you have to deal with are very extreme, and you will always see something new that you haven't seen.

  3. 16:27

    So for us, it's been about, can you think about the heuristics to be able to catch a lot of the different things as it's processed? Can you have the different engines to escalate very quickly? and ultimately we will learn new things about quality and diversity that weren't true a week ago even. And so the nimbleness to move very quickly to create and adapt your engine to catch that. What's an example of something that you would catch? Well, I think fraud is an underestimated one across data industries in general. We're ultimately creating data that doesn't exist yet versus when you're filtering through the internet. And so when you are, you're coming across very creative. People are sending generated videos or what? Yeah, very creative people where I think at one point we have folks recording YouTube videos like POV working at McDonald's and trying

  4. 17:19

    to submit that as contribution pretty much. Wow. Never underestimate how entrepreneurial people can be. That's amazing. Yeah, very much. much uh and so these have been like part of the painful lessons where yes you have to create a lot of the gates but ultimately you have to also be there where collection is happening you have to