I'm Masaki Hirokawa — artist and indie app developer.
Every May, as Google I/O approaches, I notice two versions of myself preparing in slightly different ways. The developer side is calculating: what's changing in the API this year? The artist side is looking further out: how will this technology intersect with human creativity?
This year, both have shown up again.
Twelve Years of Staying Ahead
I've been building apps independently since 2014. Wallpaper apps, relaxation apps, manifestation tools — across categories, I've crossed 50 million downloads. Through that time, watching Google's developer announcements hasn't just been a habit; it's been a survival strategy.
The thing is, the pace has shifted dramatically in the past two or three years. There used to be a week to absorb an I/O announcement, then a month or two to figure out what it meant for real products. Now, people are already shipping experiments before the keynote ends, and the next wave of news arrives before the first one has settled.
When Gemini 3.2 launched, I started thinking about what I'd need to prepare — and a week later, something else had already taken over the conversation.
I've had to remind myself: chasing everything isn't the point.
What I'm Hoping to See — Developer Perspective
The thing I most want from this year's I/O is clarity on Gemini API pricing and stability.
Gemini 2.5 Flash has been genuinely impressive for individual developers. The cost-performance ratio is realistic enough that integrating it into personal apps feels feasible. But the unpredictability in pricing tiers and free quota cycles makes the "should I put this in production?" decision harder than it needs to be. If there's any signal about long-term pricing direction, that would go a long way for people like me.
The second thing I'm watching is improvements to multimodal usability. Working on wallpaper apps for years has made me see clearly what becomes possible when image understanding gets reliable. Recommending wallpapers based on the current visual feel of a user's screen, for example — the technical pieces exist with Gemini Vision, but latency and cost keep that kind of feature out of reach for individual developers. If that changes, a lot of interesting things open up.
What I'm Wondering About — Artist Perspective
There's another way I'm approaching this year's I/O, one that's harder to articulate.
My work as an artist centers on questions about collective psychology and the structure of conscious experience — themes rooted in what I'd describe as a distinctly Japanese sense of prayer and rootedness. As AI systems become more fluent in mimicking thought and feeling, questions like what is original? and what is creativity? stop being philosophical abstractions and start touching the actual work.
Watching Gemini's development recently, I've sensed something shifting. There used to be a clear gap between "fluent" and "deep." Responses were accurate, even articulate, but they felt like echoes. Lately, I've noticed more moments where the model seems to push back — asking why I'm asking something, or offering a framing I hadn't considered rather than simply elaborating on mine.
Whether that's the beginning of something more interesting or just more sophisticated pattern-matching, I genuinely don't know. What I do know is that holding that question open — not resolving it too quickly — feels like part of what I'm supposed to be doing.
On the Art of Waiting
Both of my grandfathers were shrine carpenters, miyadaiku. One of them used to say that the act of working with your hands is its own form of devotion. Not rushing. Listening to the material. Taking the time that the work actually requires.
That sensibility — patience, attentiveness, the refusal to shortcut — has stayed with me through app development and through making art. Even as AI cycles accelerate, I find it still applies.
Whatever gets announced at I/O, my question will be the same one it always is: how does this connect to what I'm trying to make? Not how fast can I implement it, but does it deepen the work.
That's the question I'll be carrying into the keynote.
I'm genuinely looking forward to what this year brings. If you're watching I/O with your own questions running in the background, I hope this landed somewhere useful.