He had a track record that belied his years: At age 13, he had created an app for Pret a Manger, the British sandwich chain, without the company’s approval—”It was a blatant copyright violation,” he says—and it became so popular in the United Kingdom that Pret adopted it as the official app.

And so a number of human-rights groups took Browder up on his offer. He coded the first-ever app for Freedom House, the oldest democracy watchdog in the U.S., in 2013. He gave International Bridges for Justice an app that could provide training for the organization’s lawyers.

Yet despite these early accomplishments, Browder, now a 19-year-old computer science student at Stanford University, has already abandoned mobile apps in favor of something new: Bots. Indeed, his youth has tracked what is shaping up to be one of the biggest technology shifts of the decade—the move away from a dog’s breakfast of apps filling our phones and toward conversational user interfaces that live where millions of people already live: on messaging services such as WeChat, Kik, WhatsApp and Facebook Messenger.

Bots interact directly with users, serving up everything from on-demand news stories and weather and traffic updates to personalized assistance with a variety of tasks. You can already hail an Uber on Facebook’s messaging app, and now you can order flowers too. Taco Bell has TacoBot, which lets Slack users order food for pickup. Gamified conversations with fictional personas—Miss Piggy, Back to the Future‘s Doc Brown—have been done. But more complex use cases are possible.

Related: This Is the Man Responsible for Your Binge-Watching Addiction

Last year Browder created DoNotPay, a bot that appeals people’s parking tickets for free. It has saved British motorists nearly $4 million so far, obviating the need to hire a lawyer to handle these appeals. (It can also file compensation claims for delayed or canceled flights.) Like all the best bots, it improves with time, learning from its interactions with users. Now serving residents of New York City in addition to Brits, DoNotPay is one of the most noteworthy examples of a certain type of bot—one that could help build a more equitable society.

“Access to justice for the non-wealthy is a serious concern,” says Catherine Bamford, a former lawyer in Leeds who advises law firms and corporate legal departments on automation. “Legal aid budgets have been slashed in recent years. With helper bots like DoNotPay, some willing lawyers and expert programmers, legal advice could become cheap and accessible to everyone via the Internet. This is a real step in the right direction.”

The potential user bases for such helper bots are massive. Although DoNotPay has a standalone website, most bots exist on third-party chat platforms. Facebook Messenger now has some 800 million monthly active users, and WhatsApp, which Facebook also owns, has more than 900 million. Messaging apps now draw more people than social networks. And bots are easier and cheaper to make than apps.

“In the beginning, users were only using their web browsers. Then they moved from the web browser to apps. Now they are moving from apps to messaging apps,” says Eyal Pfeifel, the chief technology officer of Imperson, an Israeli bot startup. “Through a single app, like Facebook Messenger or any of the other messaging apps, they can get access to anything that they previously did with separate apps.”

Related: This Whole ‘Eating Bugs’ Thing Actually Has Legs

Thanks to advances in artificial intelligence that make it possible for bots to learn as they go and to understand human speech, we are in the midst of what Albert Wenger of Union Square Ventures has dubbed “the Great Bot Rush.” (After failing for a few months, Browder achieved his breakthrough with DoNotPay after taking a class in machine learning at Stanford.)

There is growing potential for bots to be our virtual servants or, strange as it may sound, even our friends. Last fall, as part of Universal Pictures’ social campaign for the 30th anniversary of Back to the Future, Imperson created a Doc Brown bot with its own Facebook page. You could click the Message button to kick off a real-time conversation with the bot, whose “personality” was crafted with the help of Bob Gale, who wrote the original movie.

The average length of a conversation with the virtual Doc, Pfeifel tells me, was more than 10 minutes—not bad “considering that you’re basically chatting with computer software, right?” When people realized, at the end of the 10-day campaign, that Doc Brown was about to leave for good, “they were really, really upset and disappointed.”

Imperson is in talks with various brands to develop more of these interactive “chat personas.” The most interesting and socially beneficial uses for bots, however, may lie not in entertainment, brand promotion or simple ecommerce but in leveling the playing field between citizens and governments, consumers and corporations, corporate grunts and CEOs. The startup IDAvatars, for instance, has Sophie, a kind of virtual nurse. Sophie can monitor your well-being, remind you to take prescribed medication and share information with your doctors.

“There are so many areas in the world, like medicine, where you have to pay so much for a very simple service,” Browder says. “Technology is helping to automate [these services]. And as it helps to automate them and make them free, people can have the same level of care or legal representation as almost a billionaire. That’s something really powerful that we’ve never seen before.”

One of the most well-funded bot startups is x.ai, which has created a virtual assistant, Amy, to schedule meetings. The company’s CEO and founder, Dennis Mortensen, points out that secretaries are now vanishingly rare in American corporations, reserved for the very top brass, so that even senior middle managers are left juggling their own calendars. Most people plan their own travel today, too, rather than using a travel agent. “Somehow we got fooled into believing that we were better off doing things ourselves,” he says.

Let’s say someone has asked to meet with you. Simply copy Amy or her male equivalent, Andrew, on the email and the bot takes it from there, sending the other person brief, affable messages to suggest meeting times. Once a decision has been reached, the bot puts the event on your calendar. “We see this as giving time back to the average worker that they’ve been wasting doing this very tedious task that adds very little value,” says Stefanie Syman, x.ai’s vice president of communications. In other words, it gives an average white-collar worker or freelance creative the scheduling sangfroid of a top executive.

The company raised $10 million in early 2015 on top of a $2 million seed round, and now has 64 employees working full time to improve the experience. Clearly, x.ai has gone to a lot of trouble to make Amy/Andrew seem human; during back-and-forth exchanges, the illusion is persuasively real, if imperfect. When I was setting up the interview with Mortensen, Andrew mistook a question of mine regarding a particular day and time as confirmation of that day and time, and promptly set the appointment. But the bot is getting smarter every day, Mortensen says.

Glitches and corporate missteps notwithstanding—Microsoft’s chatbot, Tay, which Twitter users manipulated into repeating offensive statements, comes to mind—the shift to bots looks to be only gaining momentum. In mid-April, at Facebook’s F8 conference, Mark Zuckerberg announced developers tools for Facebook Messenger, allowing businesses to build bots that can chat with their customers on the platform. “We think you should message a business just the way you would message a friend,” Zuckerberg said.

Although the floodgates are not open quite yet—for now, Facebook is reviewing and gradually approving bot submissions—the move could turn out to be as transformative as Steve Jobs’s introduction of the Apple App Store. Kik, for its part, launched a Bot Shop earlier this month to serve its 275 million users. Early partners include Funny Or Die and the Weather Channel.

Related: The Open-Office Concept Is Dead

The real tipping point, says Mortensen, will come when intelligent virtual agents start to talk to each other, so that, for instance, when you are planning a business trip, your meeting-scheduling bot talks to your travel bot and to your Uber bot and together they organize your flight, hotel and airport taxi pickup.

“If you want to hand over a job to an artificial agent, the best way to hand over that job is through a conversational UI,” he says. “And if you speak to Amy through natural language, what is holding back another piece of software from just speaking plain English [as well]?” That means all of a sudden all of the software we have turns compatible.” And that, in turn, will lead to greater equality: “If every piece of software speaks English, that will democratize access to it. That’s the moment where things turn sexy.”