Aaptiv, the workout app, launches a new AI-based personal trainer called Coach

After raising at least $20 million last year from the likes of Amazon, Disney and Bose for a mobile app that offered workouts led by professional, human trainers, Aaptiv is now taking a tech turn. Today, the startup is beginning a pilot of a new service it calls Aaptiv Coach, an AI-based assistant that builds personalised fitness and lifestyle plans based on a user’s goals, current fitness levels, eating habits and data input from external devices like smartwatches and fitness trackers.

Coach will mark a notable departure for the startup. Aaptiv built its name on human-led workouts produced by Aaptiv itself. (“We might the only people willing to acknowledge that we will not use AI to replace trainers,” founder and CEO Ethan Agarwal told TechCrunch last June. “The trainers create the classes and that will be always the same. That relationship and drive and the passion cannot be matched by anyone or anything.”)

Now, while the company is not removing trainers from the platform, it will be reorienting the focus of how it delivers fitness plans to users: Coach will give users an order of what should be done, with the option of using Aaptiv workouts, or alternatively following sessions led by others, or even doing the exercise on their own, using AI to measure progress.

That change, it should be noted, came with some personnel shifts at the company, too: earlier this year, the company laid off employees, including personal trainers, and while it would not confirm the exact amount, several tipsters suggested to us that it was around one-third of staff.

In an interview ahead of today, Agarwal made it clear that the reductions were not about eliminating human coaches altogether but about adding something new to the platform as it continues to scale.

Coach is being released after three years of internal development that was taking place at the same time that the company was building out the data and using it to feed and improve its algorithms. The company has to date streamed 22 million workouts across 20 countries, and all of that translates into data points that both have taught Coach, but also the startup itself about what it is that people want.

“The first phase was spent learning about how people consume classes, where and when they work out,” Agarwal said.

One thing that it learned was that while there are definitely a small and dedicated part of the population that will carve out significant time each day for workouts, there is a larger proportion that will not, and so Coach is about trying to fit fitness into those consumers’ lives.

And maybe more importantly, the rest of how an individual leads his or her life will probably give a better picture of what she or he will need to do to achieve a desired outcome (such as lose weight, or achieve a better running time, or get stronger, or whatever it happens to be).

“The idea is how can we help people achieve their goals, whatever they are?” he said. “If you work out three times a week for 45 minutes, that is basically two percent of your week. But there are a lot of companies out there competing for that two percent: gyms, in-home biking startups, and more. But the way we see it, is that the other 98 percent of your time is where your habits are formed, where you need to start to build structure.”

Aaptiv’s approach with Coach will be to create personalised daily, weekly and monthly plans for getting to a specified goal. “No one will have the same plan,” Agarwal said. For those who are already using Aaptiv, their own data will be used to start to build Coach’s profile of the person. Users will be given the option to add other sources of input, and also add in more information, to further tailor the experience.

Like many a platform, what’s interesting about Coach is how its open-ended structure could be leveraged to help Coach grow.

Today, there is already guidance on food a person should avoid or try to eat regularly, but you could imagine how one might link up the APIs from a particular appliance to inform Aaptiv’s Coach more directly on what a person is eating.

Or, you could imagine also a time when someone might even order the food she should be eating directly from the app, to be delivered by a food delivery partner.

The commerce aspects are an interesting one to consider, given how much people spend on fitness and health already today.

Take running shoes as an example: if you input into Coach what shoes you already own, and it’s tracking where you are running and how much you weigh and so on, it will be able to reasonably determine when it’s time for you to get a new pair of shoes, even if your old ones don’t look like they’re about to fall apart.

All of that will not only extend the amount of time people are spending in the Aaptiv app, but potentially present revenue streams on top of the basin subscription-based one that exists today. (It’s also perhaps a clue as to why Amazon is interested in the company.)

When I covered Aaptiv’s round last year, I wondered about how the model would scale as it expanded, built as it was on personal human connections. Coach provides one possible answer.

The potential of where and how Coach might develop is one way that Aaptiv is filling out its valuation (which last year was $200 million), and also attracting attention. From what I’ve heard, the company has been approached by multiple interested parties hoping to tap into the audience it has built and the engagement that it’s bringing to the world of health and fitness.