Adam

You were supposed to say !aside. You’ve just been charged by Devin for this tweet

When you work in fintech, you can’t always be a yes-man. I’ve learned this over and over throughout my career. At Circle I ran our yield product, which let businesses deposit USDC and get over-collateralized BTC loans. We got constant requests to increase yields by adding in more and more convoluted assets. Basically, the customer ask made sense. But oing this would have fundamentally changed our product and its risk profile. Large institutions loved the product because it was a safe way to earn. We looked into it, but the tradeoff wouldn’t have been worth it to our customers. In giving them what they wanted, we would have destroyed the returns they needed. Strong product people listen, but they don’t react. They tease out what customers really want, then make informed choices about whether to act on it. To me there’s 3 steps: • Active listening. What is the customer’s experience and perspective? • Probing with thoughtful questions. Why do they want this solution? What are they really aiming to solve? • Choosing when to execute. If we do what the customer wants, does it actually serve their needs and will it yield even greater outcomes? Sometimes, you do choose to give users what they ask for, right away. Fireblocks Flow started out as a wallet connector for banks and payment platforms, but customers wanted to use it to verify ownership and a lot more. I believe every customer ask has value. But if you respond to every ask, you don't end up with a good product. You don’t have infinite build capacity, and not every “good idea” is worth pursuing.

I’m tired of talking about the names of protocols and models. The average person couldn’t care less about Sol 6.1, Astra 6 or Astatine 85. I was really noticing this at @OpenAI Dev Day and Agentic Zero. Can people move or earn money more quickly? Solve healthcare problems? Those matter. That should be the focus of the conversation. I’d argue even the most sophisticated everyday users can’t break down the major differences between Opus 5.5, Sonnet 5.5 and Astra-5. They’d probably tell you they got a better response, but how much of that is just a feeling? Are their outcomes materially different, or are they just excited about a shiny new model name? IMO this is why personal agents are growing so quickly. They don’t bore you with the technical details, they sell what they can actually get done. This trend is most pronounced in AI, but it applies to crypto and fintech too. We’ve won when people stop talking about agentic payments and the onchain economy. Numbers go up when customers get value, not when we’re talking to ourselves about the launch behind it.

Nothing humbles you more than the angle your young kids shoot you at with your iPhone. Whelp, I guess this is just what my face looks like.

Already seeing Muse-style agents drive small businesses crazy. An owner I know, a skylight company, is getting inundated with bot voices asking for quotes. They’ve essentially had to stop taking calls. At this point, he said he’s worried even completed house call bookings may not be legit. We’re going to see a wave of companies addressing this problem. Jack is one example (withjack.com). It’s a home maintenance agent for consumers, but it functions as a platform for operators too. I think the agent-first web will be better for small biz in a lot of ways but the transition period will take some adjustment.

You mean built on wrappers of Lead Bank and CRB. Eventually without them when that middle leg is also a bank.

@Muse is going to break the internet, or many small businesses. Last week I used it to get 4 quotes in 20 minutes for work in my kitchen. As this scales, it’s going to stretch operations for businesses without agents. I think we're approaching a tipping point for how we interact with the online and physical world. You can use Muse to auto-negotiate on Facebook Marketplace. You can understand Meta’s priorities by their willingness to flood their own product with bot traffic. One downstream effect is that shitty interfaces might not matter as much. Muse will navigate that dinosaur government portal, or book your junk removal through a website built in 1999. There are pros and cons here, like with any new tech. Potentially it’s good for small businesses, but bad for the Wixes and Squarespaces of the world. And obviously, agentic shopping will change the e-commerce model. That’s not to say interfaces will no longer matter. For us at @dynamic_xyz it’s crucial. We’re handling peoples’ money so we use that UI layer to build credibility and trust. Think about the depth of thought that goes into placing a lock icon. But going forward there will be contexts where interface doesn’t matter and an agent does 90% of the interacting. The time to adapt is right about now.

Parenting young children is a series of low stakes high risk decisions 24/7. A simple answer to “did you put salt on my eggs” could mean a peaceful breakfast or an absolute meltdown that sets the day back by 2 hours. This is but one of the 100 decisions you must make with perfection before 8am..

The best founders in the world weren’t psychic. It seemed like they predicted the future, but they just made an informed bet and it paid off. @dynamic_xyz had a crazy vision: that every app would need a wallet in the next five years. It's been pretty cool to see that playing out. Again, not magic. We had a vision and we validated it. As a product leader, my lifeblood is dogfooding to get feedback as quickly as humanly possible.

If you aren’t betting on something, you're gonna lose. Gretzky said to “skate where the puck is going,” and that’s how we built @dynamic_xyz. You need to bet on a specific future, then build for it. Sometimes it makes you look crazy but it’s the only way to win.

An @NPR guy tried to get an AI doomer take from me at @OpenAI day last month. That reporter didn’t get the soundbite he wanted. Little did he know I’m the most bullish guy in the room (a room full of AI optimists nerding out). As a dad, it feels like my kids will have superpowers because of AI. And one day, they'll go to Mars.

Awesome to moderate this panel on agents and get back in the public-speaking side of my brain. xitter.gq/FireblocksHQ/status/21…

Thanks for having me. Wonderful event.

United employees earn miles for every customer rage shake.

We’re introducing Business Accounts. I’m proud of how quickly the @dynamic_xyz team is shipping and this is just the beginning. There’s a lot more coming from us. The team is cooking.

There was a time not so long ago that you could rent the bbqs along the fence there and grill up the oysters you bought from the counter with friends. I miss that experience

So when does @bendingspoons pick up retool?

Embedded wallets are in more places than you think...