Someone asked me recently if I’d ever go back to agency work. A friend from Moscow, still at a studio there, sending me their latest case study, thirty screens, beautiful gradients, a banking app they’d designed in six weeks. It looked incredible. I remembered that feeling. The rush of building something polished from nothing, handing it off, moving on.
My answer was no. And it surprised me a little how certain I was.
I spent almost two years at Spacecode in Moscow before joining VALK. Lead UX/UI, team of five or six designers, somewhere over fifty apps shipped across my time there. Kaspersky, OTPBank, fintech startups, retail, events platforms. Then I joined VALK in late 2019, one product, one designer, and now nearly four years later I’m still here, still the only designer, still working on the same platform that grew into something I couldn’t have imagined at the beginning.
These are genuinely different careers. Not just different jobs. Different ways of understanding what design even is.
What agency taught me (and it taught me a lot)
I want to be honest about this because I see product designers dismissing agency work, and it’s unfair.
Spacecode made me fast. Not reckless, actually fast in the right ways. When you’re shipping a full product every six to eight weeks, you develop instincts that most product designers never build. You learn which decisions actually matter in week one versus week three. You learn to kill your darlings quickly because there’s no time for attachment.
The visual polish I developed there still shows up in my VALK work. Working across so many industries meant exposure to dozens of design patterns I never would have seen otherwise. Banking interfaces, event apps, security software, every project had its own logic, its own user behavior, its own constraints.
Leading a team taught me something too. Reviewing junior work, running crits, helping someone understand why their hierarchy isn’t working, you understand your own craft better when you have to explain it. I mentored five or six junior designers at Spacecode. Some of them are probably running their own teams now.
The client presentations also hardened me in a useful way. You learn to defend your decisions to people who don’t speak design, who have opinions about colors, who want something “more modern” without knowing what that means. You either develop the language to handle that or you get eaten alive.
What agency didn’t teach me
Here’s what I didn’t understand when I was at Spacecode: I thought design ended at delivery.
It was built into the structure of the work. You shipped the app, wrote up the case study, moved to the next one. Whatever happened after, whether users actually used it, whether the flows made sense in production, whether the edge cases held up, that wasn’t really your problem. Not because anyone was cynical about it. Just because the project was over.
I had no idea how much of design happens after that point.
Six weeks is not enough time to understand any domain properly. I designed a banking app for OTPBank and I didn’t really understand how the back-office flows connected to what users saw. I made it look right and feel usable, but the deeper logic of the system, the compliance requirements, the actual user journeys across months of usage, I only touched the surface. The timeline didn’t allow anything more.
At Spacecode I was always the expert in the room on design. Nobody challenged me on UX decisions in any fundamental way because nobody knew more about it than I did. That felt good. It took me a while at VALK to realize how much I’d been missing by never being challenged by the product itself, by real users, real edge cases, real feedback over time.
And the client politics. That I don’t miss at all. Redesigning a section because someone’s director had a different opinion. Reverting to an earlier version a week after approval. There’s a particular kind of exhaustion that comes from designing things you don’t believe in, knowing the real reason they look that way has nothing to do with users.
What changed when I moved to product
I joined VALK in November 2019. Asset tokenization for institutional investors, the kind of domain I knew almost nothing about. In my first month I was reading about private securities and permissioned blockchains in the evenings, not because anyone asked me to, but because I couldn’t design interfaces for things I didn’t understand.
That was the first shift. Agency work doesn’t require you to become an expert in the domain. Product work absolutely does.
By the time VALK had scaled to seventy-plus financial institutions across fifteen countries, I understood the platform in a way that’s hard to explain. I knew which compliance fields caused friction. I knew which flows our institutional users took shortcuts through because the documented path was too long. I knew what “tokenization” meant not as a marketing term but as an actual technical and legal process with very specific UX requirements.
You can’t get that knowledge in six weeks. You can barely get it in six months.
The depth of product design compresses over time, not over sprints. The fourth year of working on something reveals problems you couldn’t have seen in the first year, because those problems are created by the decisions you made in the first year.
Ownership changes how you design
At Spacecode I was responsible for the design. At VALK I’m responsible for the product experience.
Sounds like a small distinction. It isn’t.
When you’re the only designer on a product and you’ll still be there six months after a feature ships, you make different decisions. You think about edge cases you’d normally cut for timeline because you know you’ll be the one fixing them later. You document things properly because future-you will need that documentation. You push back harder on product decisions because when something ships wrong, there’s nowhere to point except yourself.
The “not my problem” mindset that was almost structurally required at an agency, it genuinely doesn’t exist in product. Everything is your problem. Which is harder, but also more honest about what design actually is.
I think about the features we built in 2020 that are still running, still being used by banks in multiple countries. The interactions I designed then are being experienced by someone right now. That’s a different relationship with your work than handing off a file and writing a case study.
The loneliness is real
I don’t want to make product design sound like a straightforward upgrade.
At Spacecode I had people to challenge my work every day. Someone to catch the thing I’d stopped seeing because I’d stared at it too long. Someone trying a completely different approach while I was heads-down on mine. Creative friction is real and it makes work better.
At VALK, that friction has to come from somewhere else. From developers who catch technical impossibilities. From the CEO when something doesn’t match the product vision. From my own discipline to step back and question decisions I’ve already made. It’s a skill you have to develop deliberately when working solo, the ability to challenge yourself.
Some days it’s fine. Some days I’ll spend a week going in circles on a problem that twenty minutes with another designer would have solved.
The repetition is also real. Four years on the same product, there are moments where the novelty is completely gone. I’ve redesigned sections of this platform two or three times. The onboarding flow I’m touching now is something I’ve touched before. You have to find different sources of interest when you know a product this well. It moves from “how does this industry work” to “why is this specific interaction still creating friction after all this time.”
What actually transfers and what doesn’t
The speed of execution stayed. The ability to produce clean, considered work quickly, that’s just muscle memory and it doesn’t care whether you’re on week two of a six-week sprint or month three of a product roadmap. The visual craft transferred. The ability to defend design decisions to non-designers transferred.
What didn’t transfer: the way I thought about timelines.
My natural agency instinct was to think in terms of “done when delivered.” That took real time to unlearn. Products don’t have a delivery moment, they have a current state that will change. Every decision you make is technically provisional. That’s a genuinely different mental model and I underestimated how long it would take me to internalize it.
The “make it look good” instinct also needed recalibrating. Not removing, visual quality matters enormously, especially in fintech where trust signals are tied to perceived polish. But at an agency, looking good is the deliverable. At a product company, looking good is in service of something working right. Related, but not the same thing.
What I actually learned
Here’s something I’ve been thinking about. I learned design at Spacecode. I became a designer at VALK.
That’s not meant to diminish the agency years, they gave me a foundation I couldn’t have built any other way. The speed, the breadth, the craft. But understanding what design is for, how it sits inside a living product that real users depend on, how your decisions compound over years, that I only learned by staying.
The version of me who joined VALK in 2019 thought of design in terms of screens and flows and handoffs. The version sitting in Los Angeles in 2023 thinks in terms of systems, tradeoffs, and user relationships that extend across time.
I don’t know if that evolution was inevitable or if it required the specific circumstances, a single product, a small team, no other designer to defer to. Probably both. The constraints of being the only designer forced a kind of engagement with the product that might have been easier to avoid otherwise.
The question isn’t “how do I make this screen work.” It’s “why is this problem appearing at this point in the user’s experience and what does that tell us about the system.”
That’s the question I couldn’t ask properly at an agency. There was never enough time and never enough continuity.
Would I go back? No. Not because agency work is bad, it was genuinely formative and I respect the designers who do it well. But I’ve seen what the other side looks like, and I want to keep designing things I’ll still be accountable for a year from now.
That accountability is uncomfortable sometimes. It’s also the most honest version of what this work actually is.
Originally published on kargaev.me. Imported to blog.deeflect.com archive.