The Disappearing Wow: Thriving When Wonders Become Routine
How AI's accelerating pace reshapes product development, expectations, and why "move fast and build things" is the new mantra.
Sam Altman recently wrote an essay, The Gentle Singularity, where he observed: “This is how the singularity goes: wonders become routine, and then table stakes.”
That phrase has been echoing in my mind because it perfectly encapsulates the journey of Project Mariner over the last six months. I distinctly remember the early days, soon after the initial prototype was built. We were demoing it internally at Google, and one of the undeniable "magic moments" was when people saw the agent actually typing text into a textbox, on a website, all on its own.
Someone I spoke with recently described it perfectly: it was like the "Waymo self-driving moment, but for agents." Suddenly, we were in a world where AI agents didn't just think; they could act within our digital environment. There was something genuinely enchanting about watching Project Mariner autonomously type characters into your browser that it had conceived itself.
From Magic to "Get Back to Work"
Fast forward just a couple of weeks from those initial demos. The feedback from our Trusted Testers started to shift. We heard things like, "I want to be able to do my own work and control my browser while Project Mariner is also working." Or, put more bluntly: "As cool as it is to see an agent control my browser and type things... I’m over it. Just put it to work and let me get back to what I’m doing."
It was astonishing to witness how quickly that initial wonder faded. The functionality that initially seemed so magical rapidly became something users expected to just... work, seamlessly, in the background. The extraordinary became the new baseline expectation, almost overnight.
This brings us to where we are now, roughly six months after Project Mariner first launched as a Chrome Extension that could only tackle one task at a time while you watched. We recently released a significant update: users can now delegate up to 10 different tasks to agents, all running concurrently, freeing up the user to continue their own work without interruption.
It was a big step forward. And yet, I fully expect this version to feel "outdated" in short order. That's the nature of building in this era: the pace of innovation is accelerating at an unprecedented clip. It’s an incredibly wild and exciting time to be a product builder, but I also know it can feel overwhelming.
Adapting to the Acceleration
I'm preparing a talk for a group of PMs at Google next week about incubating new AI initiatives. I'm calling it "The Art of the Start: How to Catch, Shape, and Ship the Future." Working on these slides has given me a chance to reflect on how my own habits have changed.
For instance, I’ve made it a practice to finalize my presentation slides the night before a talk. Part of this is due to a packed schedule, yes. But more significantly, it's because things are changing so rapidly that content prepared too far in advance often feels stale by delivery time. I used to have slide decks I could reuse for months; now, my thinking and messaging require constant updates.
When I asked folks around my office for their advice on this topic, a lot of brilliant insights surfaced. But the one message ringing loudest and clearest right now? "Move fast and build things."
I’ve talked a lot recently about my 7+ AI side projects that I juggle alongside my day job. I do this because there has never been a better, more accessible time to create. And "build things" doesn't just mean traditional software applications. It means:
Create a video.
Make a cookbook (AI-assisted photos, anyone?).
Incorporate AI into your existing workflows to make them better.
Design a new personal website.
Or simply write and share your thoughts.
The tools are here, the possibilities are expanding daily, and the "wonders" of today are, indeed, fast becoming the routine of tomorrow. The key is to embrace the pace, stay curious, and keep building.
What "wonders" have you seen become routine in your own work with AI? How are you adapting?