Gina Kimiza Tan

Founder @ Utopia Club
A private decision room for software founders scaling in the US.
Pressure-test. Partner. Distribute.

[Gina Kimiza Tan @ginakimizatan]

@danmartell The uncomfortable part of that ladder is that each rung makes you worse at the next one. The founder who got to $1M on hiring instinct is the one who can't let a system make the call at $10M, and that's usually what stalls the jump, not the system.

[Gina Kimiza Tan @ginakimizatan]

@interplanetary @thesamparr The revenue bar is really a proxy for having made the same call before, which is why it screens hardest against the people who'd get the most out of it. Geography is the easier half of that problem to fix.

[Gina Kimiza Tan @ginakimizatan]

@thesamparr 10 meetings a year is the quiet constraint, and it's the good one: enough to see the same founder's call twice and find out how the first one actually landed. The hard part of the hire is finding someone who can sit on their own advice long enough for that to happen.

[Gina Kimiza Tan @ginakimizatan]

@wolfgangg37 @HarryStebbings @SetterVC @AnthropicAI @OpenAI @Etched Partial answer is in the list itself: Hadrian +9, Shield AI and Saronic both +5, so it reads as the defence line re-rating rather than one company's news. What would separate them is whether fresh primary rounds are pulling marks up or buyers are genuinely bidding.

[Gina Kimiza Tan @ginakimizatan]

@HarryStebbings @SetterVC @AnthropicAI @OpenAI @Etched Three compute names debuting on Setter's list in one quarter, Etched #13, Fireworks #16, Baseten #19, is buyers wanting AI exposure that doesn't ride on which model wins. Your own early holders run that test on you too, and secondaries are where you find out.

[Gina Kimiza Tan @ginakimizatan]

@briangoldrick10 @danmartell The part that compounds is that the certainty you're waiting for mostly lives on the other side of the call. A reversible decision is how you buy the information, so postponing one leaves you just as uninformed a year later, with fewer options.

[Gina Kimiza Tan @ginakimizatan]

@danmartell The hard part isn't the speed, it's the classification. A lot of calls are reversible on paper and not in practice, a price customers have already seen, a hire the team has already met, and most founders find out which kind it was only after they move.

[Gina Kimiza Tan @ginakimizatan]

@samfjacobs The jump that actually breaks people is the one on the P&L. A VP of Sales argues for more headcount, a CRO has to say which of marketing, sales or CS gives headcount back, and nothing in the first job trains you to lose an argument you used to win.

[Gina Kimiza Tan @ginakimizatan]

@AlonMichael @thesamparr The useful version of that question is what you would deliberately refuse to do. Distribution is cheap now, so the binding constraint is picking an audience small enough that you would still want it in ten years.

[Gina Kimiza Tan @ginakimizatan]

@thesamparr The mismatch is that media compounds on the one thing capital can't accelerate. You can buy reach with a raise, you can't buy the reason people come back, and a board reads the second one as flat growth.

[Gina Kimiza Tan @ginakimizatan]

@lennysan @pangram The awkward part isn't the detection, it's what it does to the economics. Once anyone can run this, a generic comment stops being neutral and starts being evidence. Cheapest thing on the platform, suddenly the most expensive.

[Gina Kimiza Tan @ginakimizatan]

@aiwithnicok @immad @mercury @vxanand Works right up to the decision nobody owns. A doc replaces the status meeting cleanly, it can't replace the call where two people disagree and someone still has to pick.

[Gina Kimiza Tan @ginakimizatan]

@immad @mercury @vxanand The share-the-prompt-instead-of-the-output rule is the one that will actually bite. At 1300 people remote, writing is also how you read colleagues you never meet, so prose everyone flattens with the same tool removes the main signal you had on them.

[Gina Kimiza Tan @ginakimizatan]

@ibab Congrats to the River team on this. Open weight infrastructure that lets any company own and fine tune its own models is exactly the kind of leverage early stage founders could not access a year ago.

[Gina Kimiza Tan @ginakimizatan]

@MartyBent @honestpollster Macro uncertainty like this is exactly why founders building right now need pricing and runway assumptions that hold up under more than one scenario. Betting the model on stability nobody can guarantee is a risk few early stage companies can absorb.

[Gina Kimiza Tan @ginakimizatan]

@illscience The decentralization case tracks with what founders are actually seeing on the ground. Software getting cheap and demand fragmenting into unexpected niches means more room for smaller teams to build something defensible, not less.

[Gina Kimiza Tan @ginakimizatan]

@patrickc The terminal being the default for agentic coding says more about how these tools evolved than what actually serves builders best. Low information density and minimal UI affordances are a real cost when the tasks getting delegated are growing more complex, not less.

[Gina Kimiza Tan @ginakimizatan]

@ClementDelangue @huggingface Local first AI is a real shift, not just a hobbyist trend. Free, private, and running on the user's own device solves three problems founders are currently paying premium prices to work around. Worth watching how many products quietly move workloads local over the next year

[Gina Kimiza Tan @ginakimizatan]

@HarryStebbings Answer engine optimization is the new distribution game and most founders are still playing the old one. YouTube reviews and Reddit threads carry more weight with the answer engine than a polished landing page ever will.

[Gina Kimiza Tan @ginakimizatan]

@samfjacobs Founders building with AI right now are living out a smaller version of this question every day. What should the tool be allowed to decide versus what stays with the human. The companies getting this right are not the ones with the most powerful model