Startups Challenge Big Tech’s Hold on AI Personal Assistants

Designing intelligent personal assistants isn’t just a game for tech-sector titans anymore.

As The Verge reports, API, open-source tools and the ability to rent computing power are making it possible for more and more startups to leverage academic and industry developments — like natural language processing — to make apps tightly focused on a particular task.

Why the Barrier Dropped

Each of those three ingredients removed a cost that used to keep small teams out. APIs let a startup plug in a service, such as speech recognition, that someone else has already built. Open-source tools give developers free access to code and research that once lived inside big labs. And renting computing power from a cloud provider means a startup doesn’t need its own data center to train or run its software.

Put together, that changed the strategy. A startup couldn’t hope to build a general assistant that did everything as well as Siri or Google. But it could pick one job, like booking meetings or tracking trips, and try to do that one job better.

Three Startups, Three Bets

X.ai is a startup based in New York City that’s raised almost US$35 million since starting in 2014. It makes a scheduling assistant, Amy Ingram. When ‘she’ is copied on an email, the software coordinates a meeting day and time.

EasilyDo, another AI-assistant startup, is specifically tailored to help with travel — an area already dominated by Apple’s Wallet and Gmail.

EasilyDo says its solution recognizes more travel companies than its biggest competitors, leading The Verge to ponder, “Does it matter if a startup is marginally better than Google or anyone else in a domain, when the bigger player is good enough?”

That question hasn’t stopped Viv, a startup founded by Dag Kittlaus, the maker of Siri. Viv’s AI can write its own programs, instead of relying on human coding. Kittlaus calls it a breakthrough in computer science: “It’s going to change the way programmers work with computers.”

Whether downloaded apps can replace baked-in personal assistants remains to be seen.

One thing’s for certain, though. As screen fatigue grows and more people seek out other ways of interacting with technology, intelligent personal assistants — whether it’s just a voice inside your smartphone or an actual robot — will be the future.

What Happened Next

The Verge’s question turned out to be the right one to ask, because two of these startups ended up inside bigger companies.

Viv went first. In October 2016, only a few months after this article, Samsung announced it had agreed to acquire Viv Labs, describing Viv as an open AI platform that lets third-party developers build conversational assistants. In March 2017, Samsung introduced its own assistant, Bixby.

X.ai followed in 2021, when event software company Bizzabo acquired x.ai to add AI-powered scheduling to its event platform. Amy Ingram’s job, finding a time that suits everyone, became a feature of a larger product rather than a standalone service.

That pattern is common for focused AI startups. A narrow assistant can be genuinely better at its task, but it’s also exactly the kind of capability a bigger company would rather buy than build.

The Big Assistants Fight Back

The “baked-in” assistants the article mentioned didn’t stand still either. The arrival of ChatGPT, first released in November 2022, reset expectations for what an assistant could understand and do. The large platforms moved to catch up.

In June 2024, Apple announced it was integrating ChatGPT into its software, saying Siri could tap into ChatGPT’s expertise when helpful, with users asked before any questions are sent. In February 2025, Amazon introduced Alexa+, a new version of its assistant powered by generative AI.

In other words, the built-in assistants are now borrowing the same kind of general-purpose AI that startups once used to compete with them. That makes The Verge’s 2016 question even sharper: when the default assistant on every phone can handle scheduling and travel reasonably well, a specialist has to be a lot more than “marginally better.”

What It Means for AI Founders

For founders building assistants today, the lesson from Viv and x.ai is to be honest about where you’ll end up. If your product does one thing well, ask whether it can become a business on its own, or whether it’s more likely to be a feature someone else buys. Both can be good outcomes, but they call for different plans.

And build around something the platforms can’t easily copy, such as deep knowledge of one industry, data your customers trust you with, or a workflow that fits how a particular group of people already works. A clever model alone isn’t enough when the biggest companies in the world are building similar capabilities into the devices people already own.