College students are supposed to arrive ready to learn, manage themselves and make consequential decisions. In reality, many are walking into adulthood with battered attention, uneven executive function and support systems built to react after things go wrong.

Jamie Wood thinks that is a product problem. In part three, we look at AUTONOMIC's 18-to-24-year-old beachhead, early working-memory results, the gap between a buyer and a user, and how an AI coach might make useful brain science legible without turning it into another institutional lecture.

Transcript

Jamie: Right now our beachhead that we're really going after is that 18-to-24-year-old demographic—college and university. But they don't have stress. High cognitive need, high demand environment, that's our sweet spot. What's interesting about that though is that you can still lay down behaviors easier. They have intrinsic motivation to do well to have their brain firing. As well as probably the extrinsic, which is whether it's a scholarship, their parents, what have you. When we think about what does it mean to help somebody in that high cognitive time of need, give them a toolkit and help build neural pathways that they can then take into their careers, what we hope to happen is the next time they come up against a cognitive limitation, whatever it may be, first child, first home, divorce, and they're noticing, "Hey, listen, something's not great with my brain"- we wanna be their next thought. Their next thought. That demographic is really important to us for a number of reasons. But also to train the models. Who's paying? Colleges and universities. This falls under student success.

Pablos: Student success is a thing they pay for?

Jamie: Students probably don't realize this, but every semester there's a student success fee that comes out of their tuition. They have to pay it. A line item. Student success is to do with retention, passing grades, services that help them succeed in that environment. The colleges and universities very much care about their student success budget. Can't just spend it on booze and drugs. Or yoga classes, gym passes. That's student wellness. Different. Student success is literally about: how is this student succeeding academically? Autonomic, a proactive solution—that's where we're plugging into. What's interesting about that? Really large data sets.

Pablos: You can get a lot of students.

Jamie: Colleges right now that we're working with collectively have 147,000 students. It's three colleges. We're doing pilots in each.

Pablos: Let's start with the administrators. I'm an administrator at some of these schools. I got this budget. I gotta spend it on something. My options are what?

Jamie: There's counseling. There's career prep.

Pablos: The other things in that bucket could be Autonomic, where it's like, "Hey, you just need to chill out. Here's an app that will help you."

Jamie: You just need to train the right behaviors to excel. This environment isn't changing. You need to learn how to excel in it. Adapt to it, and you need to train those behaviors. 'Cause if we go back to the fundamentals of how we looked at building all this, high-demand, fatigued brain. What is the brain trying to do? It's trying to conserve energy, and it's trying to favor habit loops that require less willpower. Even if somebody says, "Hey, you should really do—it's working for me." And let's say that individual goes, "I'd like to."

When they're fatigued, it's difficult for them to lay down that behavior, which is where that coach piece comes in. You gotta hit all of those things. To have a learner learning the right thing at the right time that's applicable to them. It has to be quick. It has to adapt to them over time, iterate with them. It has to provide insights, accountability. All those things need to hit in a quick way to help this person.

Pablos: Do they get a high score?

Jamie: The gamification piece is interesting. We haven't built anything in yet, but it is a very big topic of discussion, 'cause there's a lot you could do. They win Bitcoin? I bet they'd love to. Oh my gosh. Do you have some to share? If you look in the student success budget, maybe you'll find some.

Pablos: From each student's perspective, this is a service that's available to me. I try it out. Do you have metrics on what retention is like, or how successful this is, or how much better I can perform with it?

Jamie: Absolutely. What's the best case? We've done behavioral studies, looking at the statistics over time of people reporting their perceived stress levels, sleep energy, mood, focus. Then we also have, objective efficacy study where we did brain scans and measured people's brains, had an intervention group, control group. Intervention group got Autonomic, control group didn't. I will say preliminary studies, which is why we're now moving forward with more grants and raising money. But what we saw that was really exciting is we saw an 11.6% increase in working memory, which is huge.

The brain retains information, problem-solves, makes decisions. We reach statistical significance with that. We saw an improvement in focus by about 9%, a reduction in the level of fatigue in the brain by about 4.5%. We're a very science first company, so we wanna prove efficacy because ultimately we are in the brain training industry which is full of bogus garbage, things that don't work. But that industry is growing significantly, and it's because people are looking for solutions.

To actually be a company that's highly efficacious, proven to work, and comes with all the things we're talking about. There's a really highly efficacious and proven to work app that's going on fire sale right now. TikTok. You guys could probably just buy that up and then train brains to do better things than the current version, which is training brains to do some pretty dumb shit. It's interesting. I always think: God, we have all these platforms that train the behavior of distraction. What would happen if we had a platform used at scale that actually trained the behaviors of enhanced cognitive function? That's the version of TikTok. We're not allowed to use it. But their kids are fucking thriving. That's what I hear. The status of the company now is you're doing these pilots with schools. Presuming that goes well, go sell to more schools.

Pablos: Keep doing R&D to try and make the product better. Maybe eventually the finance industry will come around and realize that their people should use this too.

Jamie: We bootstrapped the company up until this point. We didn't start building Autonomic proper until about four years ago. At the beginning of 2020 was the shift. My experience prior to that wasn't in building this technology or biotech company. What we're looking to do now is advance the AI capabilities and translate from a web app to a native app to have better data capture and better user interface.

We're seeing huge success with the web app, but it's clunky. What do you need?

Pablos: You need some nerds to help you build mobile.

Jamie: I need some awesome, unique weirdos to come work with me.

Pablos: That's great, because that's who listens to this podcast.

You have ambition to train some kind of machine learning model on the data you're getting from all these sources.

Jamie: If you get lots of students with lots of data, it could be interesting to train a model so you can provide better recommendations based on the data coming back. Our data set—people that have been through or currently on the platform—represents about 1,000 people. Two models have already been trained on that data set. The model behaves in the same way a brain performance expert would: recommend a habit based on the data that's come in, and recommend the right next survey. Where we need to go is: take all of the data and interactions from the human in the loop to train that model—the interaction of the coaching model. Put those three together: reinforcement. Somebody listening should be able to do that. Call Autonomic.

Pablos: This is great. You're bootstrapped to date. You got some grant money.

Jamie: Convertible debt note. Got some investors.

Hopefully get revenue from the schools soon. All of the pilots have been paid. Even from our alpha beta, everybody always paid to use the platform. We never gave it away for free. That's really impressive. They give a lot away for free, and they run out of business. But I think the demand is there. We need to scale to meet it.

We're doing these pilots, but we max out at 300–500 students per pilot. Why? What's the breaking point? The breaking point is the human in the loop. I could add more humans in the loop, but then we're starting to weaken the signal of what we're trying to do with training the model. I have the perspective that more humans can create more variations of the model, which I know might sound like what you want, but we need to prove out this first phase to then expand out.

I look at it as a very niche, almost domain-specific. There shouldn't be a lot of variations to how that coach interacts with an individual when it comes to improving brain performance. As a perfect example, part of the experience for the user is that they reflect. When they answer yes or no to habits, they reflect on why they've rated it this way.

As a recent example, somebody talked about they're having a hard time because they lost their father and they're grieving. What can we do in terms of tying what that person's reflected on to neuroscience? The keyword is grief. They can receive information: some neuroscientists view grief as a temporary brain injury based on what happens. It's always explaining the neuroscience of why. But you don't need to go broad in that. How we've designed it is first and foremost how the AI coach needs to speak to somebody, you need to make sure they're heard, understood, and supported, then coach them towards a behavior. You would want the model to respond correctly to what that person said. But the hard science is grief—the trigger word providing a template around that—and that iterates. Somebody else talks about, "The new Fortnite came out, and I ended up staying up really late, but I didn't feel good after." Perfect. We know Fortnite is a video game.

You can talk about the specific impact of Fortnite on the brain, which there is research around: that information share.

Pablos: That seems good. Now is a good time.

Jamie: Have an LLM distill these papers into something digestible—probably manga, or whatever people can handle.

Links

Recorded on January 16, 2025