#54: The Human Data Engine Training Smarter AI Models | Micro1 CEO

**Ali Ansari** (0:00)
When models get very good at the, let's call it the current action space of all things in law or any other domain, that means that humans in those functions will be able to use models really nicely for their job, which means that their job sort of like action space will change, and they will rely on models a lot for everything they do. So they'll have free time to do sort of higher level, more nuanced tasks.

**Derek Andersen** (0:38)
This episode is brought to you by Google for Startups.
Welcome back to Divot. Today we're in downtown San Francisco with Ali Ansari, founder and CEO of Micro1, company building the human infrastructure behind the next era of AI. Micro1 is being reported as one of the breakout players in AI data, with Reuters reporting a multi-hundred million dollar valuation. A bigger story is Ali and the team's vision, connecting specialized human expertise with the AI labs and companies shaping the future of work. Today, we're talking about how the large language AI models train their products, where the data comes from, and how it continues to get smarter over time. And finally, why human knowledge will continue to be the driver of AI. Hope you enjoy the conversation. We're like right in the heart of San Francisco, right in the epicenter, the ground zero of where everything in AI is exploding right now.
Tell me a little bit about being in it, being one of the players in this world. Like what is going on? Describe it to someone who isn't spending all their time in San Francisco right now.

**Ali Ansari** (1:44)
I think it's really fun and overwhelming to be in it. There's sort of two very broad categories of things that are happening. One is folks that are training models, and two is folks that are figuring out how to use models and implement the models. And we're sort of in both of those in a way with what we do at Micro1, which is we help the foundational model companies improve their capabilities and we also help enterprises evaluate their way into making real agents like production level agents.
So, yeah, I mean, I think it's definitely a bit of a bubble in SF that we sort of assume that the rest of the world sees and feels and thinks the same way in some cases. But I think at the same time, there's also this kind of broader set of folks that are pretty directly actually contributing to AI as well, which is what we were speaking about earlier around this notion of the hundreds of thousands of AI trainers that are around the world that are giving their sort of structured judgments to directly improve molecule abilities. So I think because of that, it's actually a very much more broad, I would say understanding of what's happening because of those roles specifically. I think it's a good thing for folks that are deeply involved in building up.

**Derek Andersen** (3:10)
Just describe a typical day for you. I know a lot of your team is remote. We're here in your office in San Francisco. But what is the pace of work? Are you all day, all night? Are you on your phone late at night? Are you here in the office with your team? Is it, are you at events? Are you going to your customers and other people?
What's happening on a daily basis?

**Ali Ansari** (3:37)
Wake up around 8, 8.30, somewhere around there. Start the day with a bunch of calls, back to back till maybe one or two. I try to limit it as much as possible, but it ends up being usually until the afternoon. Finish those, and then a lot of it is research, syncs with the team, engineering calls, customer meetings, going in person or popping on Zoom. And then after roughly one or two PM, it is a huge amount of going through Slack and just responding to hundreds of messages, which often ends up taking a good portion of the day, which is a bunch of small actions that keep things going forward. And then around six or seven is when I try to shut down Slack a bit, but oftentimes it doesn't entirely work, but try to put it away, do a quick gym break, which is non-negotiable for me, or else I would go a bit insane.
And then after that, I usually either come back to the office or go work the rest of the night from home. And from like 8 p.m. all the way to roughly 1 a.m. or so, is more of the focused work, where I do product reviews, I check a bunch of big files, review designs, try to set up some new pipelines for customers, things like this, which is the fun part of the day, but the rest is like keeps the company moving forward.