Sarthak interviewed me at Data Hack Summit 2026. Here’s the Q&A.

Interviews and panel discussions force me to think quickly and tend to crystallize thoughts I haven’t yet fully processed. I really enjoy them, and this was no exception.

Q: Any key takeaways you’d be taking home?

  • Knowledge graphs may not be as hot a topic as people might think, that’s definitely one.
  • Forward-deployed engineers have not yet caught on as much as I would have thought they had.
  • Surprisingly, no one mentioned outcome-based pricing as a theme, which is interesting; I’m guessing that there’s still a lot of pressure against it.
  • And loop engineering does seem to have potential, but the verdict is out.

Q: Is judgment now more valuable than analysis?

  • Some parts of judgment are, and the rest of it is as useless.
  • Judgment you can explain is something that you can train an agent on. So that’s not the part that’s going to be useful.
  • Maybe some aspects of judgment are about intuition, about noticing something and saying, “Oh, this could be the next trend,” which (A) happens to work and (B) is not something that is easily teachable. That will become more important.
  • But how can you teach judgment if that is not teachable? Unfortunately, I guess people just have to imbibe it in some shape or form.

Q: How do you spot a truly important insight?

  • I used to evaluate whether it was big, useful, and surprising - BUS is the acronym that I use.
  • I’ve taught agents to do that using a skill; they do a fantastic job of it.
  • So if your question is how do you spot an insight that an agent can’t, then my answer is I don’t know.

Q: How do you filter through the noise fast?

  • Agents can help if we know how to ask better questions.
  • E.g. If we know to ask, “Are the smart people following this trend?” or “Are we seeing any early trends?”
  • The book Superforecasting describes ways in which forecasters are able to capture that signal.
  • But with agents, I think we can be superpowered, making a regular forecaster a superforecaster.
  • It’s not something that I’ve personally tried, but now that you mention it, I’m going to try.

Q: When would you publish an AI-found insight without changing it?

  • I’m just trying to think how many I already have. And the short answer is several, so I can’t remember when I first did it.
  • Already done.