Evidence receipt / commitment
Published · transcript-backedRahul Sonwalkar: commitment
9 Mar 2024 Latent Space Top 5 Research Trends + OpenAI Sora, Google Gemini, Groq Math (Jan-Feb 2024 Audio Recap) + Latent Space Anniversary with Lindy.ai, RWKV, Pixee, Julius.ai, Listener Q&A!
“If you've seen that code interpreter, that's pretty hard for users to do. So we focus on data and that use case, and we will do that.”
Source trail
Everything needed to verify it.
- Speaker
- Rahul Sonwalkar
- Attribution
- Verified speaker
- Claim type
- commitment
- Recorded
- 9 Mar 2024
- Publisher
- Latent Space
Transcript context
…What's like the practical difference in UX between this and just trajectory code interpreter? Great question. Yeah, the question was, what is the difference between Julius and code interpreter? Really, there isn't. It's just better. We're focused, we're focused With people, or people who do stuff with data multiple times a day. And we talked to a lot of these people, and we said, Okay, how can we build things for you that would help you do your job? So, an example of this is on chat. gt, often times they'll give it a data set. People try to write their code, and sometimes that code has errors. And it kind of goes into this loop of trying to fix these little errors. What we have focused on is, okay, how do we prevent that from happening? So we looked at thousands of users using us daily. Collected data on where these errors happened. And focused really hard on fixing those errors. Beforehand, before they actually happen at runtime. This could mean a bunch of rules. This could mean, you know, prompting changes, et cetera, and just preventing that from happening. Second of all, we have features that allow people who do stuff with data on a daily basis to go deep and do the last mile of analysis done. That could mean, you know, You can click, show code, go into the code, edit the code changes. You can also give natural language instructions on the code. Finally, let's say you have this graph. And I want the graph to have some changes. Like, I want it to be a bar chart instead of instead of instead of a line graph. You can kind of just go in here and give natural language instructions to let the user take what the AI has done for it and then take it to the, to the finish line. If you've seen that code interpreter, that's pretty hard for users to do. So we focus on data and that use case, and we will do that. Cool thanks guys! That's unfortunately all the time we had to feature demos, but many thanks to Botpress, Markov, Kura. ai, Sweep, and Motif as well for being finalists. For the last part of our anniversary celebration, we wanted to turn over the mics to you, our dear listeners. We hear so many great stories from listeners about how latent space has come into their lives, and we've never had the opportunity to feature them on the pod till now. Our first listener is Balaz Nemethy from Hungary, who talked about one of the most delightful gems in the latent space community, our weekly Discord paper club.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.