I joined Imprint in April. Before I joined, Dan Na, my hiring manager, walked me through what a credit card launch actually looked like. The long hours, the controlled chaos, the everything-at-once nature of turning on a card for hundreds of thousands of people in an instant. It was equal parts warning and pitch.
I had that warning when I agreed to join so I knew that I’d be doing a lot of work, moving extremely quickly, and learning a ton in the process. What I didn't anticipate was where the speed would actually come from. It turns out velocity here isn't primarily a function of pressure, it's a function of how many obstacles exist between noticing a problem and fixing it.
Question everything
One piece of advice I got on my first day: don't assume anything. It's a startup, and things that feel like table stakes sometimes got cut somewhere along the way or just never made it onto anyone's radar.
That landed for me pretty quickly. A few weeks in I noticed our transactional emails were coming from “no-reply” instead of “Imprint” and had no logo when viewed in Gmail. Alongside other banking emails it just looked bad, and it trains customers to trust emails from sketchy senders. In some orgs you’d assume surely that’s on purpose, the lift must be high, or there’s a technical blocker. Because of the warning I had gotten, I assumed it just had not ever been the priority. The second piece of advice I got that day: if you see something that bugs you, fix it. So I did.
In many organizations, noticing a problem is the easy part. Figuring out who owns it, getting it prioritized, and finding a slot on a roadmap is where months disappear. Here, the approach was simpler: if you see something that should be better and you can improve it, do it.
The autonomy
A few weeks later I tried to add my Imprint card to Monarch Money (a financial aggregator I’ve been using since Mint was shut down by Intuit) and quickly realized it wasn’t going to work. Financial aggregators aren’t supported by the login system we use. For a credit card product, that's a real gap in the customer experience, but it just hasn’t been an organizational priority to this point.
You may have seen this meme
That's the kind of passion I have about tracking my finances in an app. The only difference is I'm not quitting after I fix it. So I wrote an RFC to implement the authentication and data APIs required. It was approved in an hour. Now I'm building it, running in a background Claude Code session while I'm doing everything else.
That pattern goes all the way to the top. Will Larson, our CTO, is shipping more features by himself than many ICs I've worked with and somehow still has deeper context on every running project than you'd expect. Leaders at Imprint set direction but they don't disappear into strategy docs afterward. Everyone is in the work, and waiting for permission to act on something you know is right starts to feel like a strange choice.
AI-native
This one is hard to describe without it sounding like every other company that added "AI" to their all-hands deck. It's different in practice.
Our main support channel has every incoming question auto-answered by a bot. Our onboarding doc was rewritten so that instead of a checklist, you point your agent at it and it sets up your environment. We're building internal tooling to let engineers share their coding agent sessions, not just the output but the strategies behind them, so what one person figures out everyone benefits from. At any given moment there are probably 5 Claude Code sessions churning away per engineer. The work I talked about earlier to allow Monarch to connect is just one of my 5 background tabs. The incremental cost of running a 4th, 5th, 6th Claude instance once you are used to the context-switching and have a good harness set up is so close to zero there’s virtually no reason not to do it.
The expectation, backed by actual tooling and budget, is that you keep finding ways to do more with it. It doesn’t stop at tooling though, you have to be willing to fully reshape your processes. It's just how work gets done here.
The velocity
The pace at Imprint is real, but it doesn't come from pressure. It comes from an environment where the right conditions are present. Flat enough to act, trusted enough that you will, equipped enough to move fast when you do. None of this works without trust. People are expected to make decisions, move quickly, and occasionally be wrong. The tradeoff seems worth it. If you get good at identifying the really important and potentially costly decisions vs the ones that are easy to undo, you can suddenly move faster with lower risk.
I never expected that working at an insanely fast pace could actually feel this good.
Top comments (0)