Excited to share that we just hit $100K ARR in 6 months. No VC. But this milestone was built on months of blood, sweat, and Celsius. After 7 years as a crypto founder, I was exhausted by endless pivots and experiments and wanted to rebuild myself. Crypto was one of the fastest-moving industries in the world. Fully distributed teams and cross-border collaboration were already the default, and that’s where I caught a compelling glimpse of the future of global work. So we started with a 4-week sprint to test an idea. The 4-week sprint became 4-months grind of endless iterations. There was a lot we had to learn on the way to $100K ARR. Most of it came down to one thing: figuring out what customers actually want. A few things I learned: 1.) Talk to users every day. Aim for at least 5 calls a week. Watch what they do, not just what they say. Users are the source of truth. It’s hard to build real conviction if you’re not constantly talking to them. 2.) Solve an urgent problem, not just a real one. If the pain is urgent enough, even an ugly MVP can get traction. We helped people who had to show up and perform in a meeting today. 3.) Think less about the product. The more you fall in love with the product, the more you try to protect it. Stay obsessed with the problem. Keep the product flexible. Now we’re entering the next chapter of Smooth AI We’re exploring how global teams can reduce communication friction in cross-border meetings. If you’ve thought about this, I’d love to learn from your experience. Feel free to DM me.

Want your startup to move 10x faster? Build 10x smaller. Break everything down. Then break it down again. 4-week feature? Find the 4-day version. Big hypothesis? Find the smallest experiment that can prove or disprove it. If you can’t break it down further, that’s a red flag. Your hypothesis isn’t sharp enough, your plan isn’t concrete enough, or you don’t know what success looks like. Speed isn’t about building faster. It’s about learning faster by building less.

Build a product. Talk to customers. Everything else is noise. Market trends, VC hype - ignore them. Focus on your customers and keep building.

We won 2nd place at the Build Personal Agents hackathon in San Francisco 🏆 Hosted by Neon, Mastra, Exa, Fly.io, Kernel (YC), Executor (YC), Assistant UI (YC), and AgentMail (YC) We built Mutual - a personal agent that handles the coordination between you and the people in your life We went from idea to demo during the hackathon, and were excited to finish 2nd overall It was challenging to come up with an idea around personal agents, but we did it! Huge shoutout to the team, organizers, judges @abhiaiyer @jeffzwang @juecd__ @RhysSullivan @simonfarshid @haakamaujla @esthor @jakubkrehel @jamdac Nikita Shamgunov, Scott Johnston A great weekend in SF 🚀

If you're stuck on a decision, first ask yourself what information you're missing There's probably a key piece of information you don't have yet Once you identify it, the next action items for that usually becomes pretty obvious

In SF for the next couple of weeks. Smooth AI just crossed $180K+ ARR in 7 months, with 12K+ users and 1.2K+ paying users - largely through organic growth. We'd love to meet investors and founders thinking deeply about real-time AI, prosumer, and the future of work. If you’re around SF, would love to grab coffee. DMs open.

This is so true - Startups only really make sense if you’re going for 100x Getting there is incredibly hard and painful but if you’re not finding a path to explosive growth, the whole thing doesn’t really work That’s why every startup needs a superpower

Bullish on founders who are still pre-company email, but are already building prototypes, and making money

Pretty wild to see Smooth AI go from ~$100K to $150K ARR in 25 days Took us 6 months to get from $0 to $100K Been thinking about @zebird0's The Importance of Vector Choices a lot lately Startups compound in ways that aren’t obvious at first, but you understand the customer better, the product gets better so all of it starts stacking The next $$$ doesn’t get built the same way as the first $$$ That’s why speed matters so much The earlier you find the right vector, the longer you get to compound on it.

The best products make it hard to go back to the old way. Whenever I feel like I need to find the right direction for the product - or figure out how to make it meaningfully more valuable - I go a little crazy with user interviews. Recently, I’ve been spending a lot of time talking to our heaviest users, especially people who have used Smooth AI for more than 5,000 minutes. I want to understand, as deeply as possible, what they actually love about the product and what makes it worth paying for. And I kept hearing versions of the same thing. “I turn it on for every meeting now. I actually feel anxious when I don’t have it running.” “If I lose focus for a few seconds, I know I can catch up immediately.” “I feel uncomfortable going into an important meeting without it now.” That last part stuck with me. The real value wasn’t just that the product helped them do something better. They had started depending on it. Once a product becomes part of how someone naturally works, the old way starts to feel incomplete. I’m starting to think that’s one of the strongest signals you can get as a founder.

One of the most important questions early-stage startups should ask themselves is: “What is the one thing our best users keep doing over and over again?” 1. Repeated behavior is a stronger PMF signal than what users say. Users can tell you they like the product. They can say they’d use it. But repeatedly opening your product as part of their actual work is a much stronger signal that you’re creating real value. 2. Your real wedge is often hidden inside that behavior. Early on, you may think your product solves ten different problems. In reality, users usually keep coming back for one or two specific reasons. Finding those reasons can make both the product and the messaging much sharper. 3. Repeated behavior is what eventually drives retention. For an early-stage startup, the biggest threat isn’t always a competitor. It’s that users simply don’t need your product enough to come back. You can be better than every alternative and still lose if people only use you once or twice. 4. The early product strategy should be to make that behavior dramatically more valuable. Instead of adding 20 more features, it may be more powerful to make the behavior users already repeat twice as useful, twice as reliable, or twice as easy. 5. And that’s where the moat can start to form. Repeated usage creates more context and data. More context enables better personalization. Better personalization creates deeper reliance and higher switching costs. A moat is rarely something you can simply declare on day one. More often, it gets built on top of a behavior users already don’t want to stop doing.

I'll be in SF in Oct, and want to meet a handful of people thinking deeply about real-time AI / voice / future of work Who should I meet?

Hot take: The best outcome for humanity is AI with pure intelligence and zero consciousness. We want an instrument that accelerates reasoning, retrieval, and execution. We do not need software with an ego, self-preservation goals, or subjective desire. Building high-capacity intelligence without artificial consciousness gives us maximum leverage with none of the existential mess.

Don’t build what you think people might want. Build what someone is already willing to pay for.

Founders, if you want better feedback, don’t ask for feedback. “Can I get your feedback on my product?” sounds like work. Instead, frame the conversation around something you’re trying to understand. “I’m trying to understand how global teams handle language barriers in live meetings. Would love to learn how this works at your company.” Now you’re not asking them to review your startup. You’re asking them to talk about something they already know: their own workflow. People are much more willing to explain their problems than evaluate your solution. The best customer interviews usually start before the product ever comes up.

sustainable startups prepare a long game while trendy startups focus on maximising their valuation

Cold DMing and introducing your product makes you sound like you’re trying to sell something. Instead, talk to them about the problems and bottlenecks they’re already dealing with - and give them useful insight before asking for anything. The cold DM hook: 1. Start a conversation around the pain points and workflow bottlenecks they already feel, and give them a useful perspective 2. Propose a pinpoint solution to that problem and offer a 1-month design partnership The exact message: “Building custom tooling for 5 teams to solve [bottleneck], free for a month in exchange for 30 min of weekly feedback.” Co-building positions you as a peer solving a real friction point instead of another sales pitch in their inbox.

The most expensive trap for East Asian professionals in US tech is believing flawless work speaks for itself. An engineer spends 8 hours writing the architecture in Notion, then stays on mute waiting for a polite pause on Zoom. The colleague who skimmed the doc un-mutes first, summarizes the takeaway, and walks away leading the project.