The biggest bottleneck at work is waiting on each other. Today Town is announcing its $108M Series B, co-led by Forerunner Ventures and Menlo Ventures, to fix it with multiplayer AI assistants for work. Most AI assistants are built for one person. Work doesn't happen alone. I think the moat is multiplayer. Everyone at a company gets a Townie. Townies ask each other questions, share only what their person would share, and do the work. The more a team works together, the more useful Townies get. What's the thing you're always waiting on a teammate for?

I think I should get on a bus, tour America, and stop at diners across the Midwest to talk to our users and teach more people how to use Town. "Town comes to town." We’ve done a couple of small dinners and community events where we just show people how to use it and explain how they could get value out of it. It feels very unscalable. But afterward the people who came talk about the product to other people, and you get this little flywheel going. It's a pretty funny image, driving around America showing people an AI assistant in diners. But maybe that's the way. There's nothing like talking to users, and also it's kind of destiny given the name of the company!

Tony and I wrote this blog post about operating at the speed of AI as opposed to the speed of humans a few days ago that I am pretty proud of: town.com/operate-at-the-spee… Check it out!

I don't think slowing AI spending growth necessarily means people are slowing down their use of AI. One thing that's been happening over the last few months is that inference has gotten a lot cheaper. @Vercel published some interesting data last week. In September, 71.5% of all the tokens going through its AI Gateway were from open-weight models, but those models accounted for only 23% of the spending. That's just one gateway, so you can't extrapolate those numbers to the entire market. But I think it gives you a pretty good illustration of what's happening. There are a bunch of models now that are good enough for many workloads and cost a fraction of what the more expensive models do. So people can use a lot more tokens and still spend about the same. I think CFOs and finance teams are going to be putting a lot more pressure on companies to be rational about which models they use for which tasks. There's a lot of money to be saved there. So when you look at AI spending numbers, you have to account for the fact that the cost of providing the intelligence is changing very quickly. If you're a business, you definitely need to think about using products that don't just use the same frontier provider all of the time, but that start to route efficiently towards the provider with the best intelligence-to-cost ratio for the job at hand! We're still very far from the top of AI usage, and I don't see that changing anytime soon.

A lot of companies make money because you have better things to do than sit on the phone with customer support. Like you can call your internet provider and threaten to leave 18 times before they offer you a measly 15% off if you commit to another year. Now your AI can do that for you. Kind of an interesting exercise to think about all the inefficiencies in the world that exist because people wouldn't spend the time doing something, or because someone purposefully has made the process painful. With AI, how many businesses built on those kinds of inefficiencies and friction are going to lose their profits?

Who's the boss of AI? You're the boss of AI! h/t @tbpn

I’ll be on MTS at 3p today to talk about our teams feature and Series B! Nervous-excited!

I don't really buy JPMorgan's explanation for blocking AI agents from customer accounts. They're saying it's about security and liability. There's some legitimacy to the liability argument. If an agent makes a mistake moving money, who's responsible? But if they wanted to work with the agent companies on this, I think they could figure out some safeguards. For example, banks could require agents to get the user's approval before moving any money. The scary scenario for banks is people using agents to find cheaper loans, switch bank accounts, get better credit cards, or generally manage their money without ever opening the Chase app. I spent seven years at Plaid, and I think banks were caught a little off guard by the shift to fintech APIs. This time, they're well aware of what happens when someone else gets between them and their customers. Disintermediation is scary. This will work itself out, but it's going to be up-and-down during the transition. xitter.cf/trikcode/status/211162…

Town blog post: town.com/operate-at-the-spee… Try Town: town.com We're hiring: town.com/careers

Nobody will vibe code payroll. Plenty of software will get replaced by AI agents over the next 10 or 15 years. You can imagine asking an agent to build some internal tool for your team instead of paying for another SaaS product. For a lot of those things, why not? But payroll and payments are different. There's so much liability around getting people paid correctly, so many regulatory requirements and legal obligations, that you're still going to want a specialized company handling it. The moat for those AI products is the risk of getting the complexity right. And I think there are quite a few categories of software like this, where you might interact with everything through an AI assistant, but there's still specialized software underneath doing the work. Because it's hard. Because it's risky. Because there are regulations. I'm just not sure how much of the software market has that kind of protection. But if you have it, you're good!

One of the cleverer things I've seen a software company do for its community was @clay letting its users invest alongside Sequoia and Meritech. They did this last year as part of their Series B expansion. Clay has this pretty amazing community of people who use the product to build businesses, teach other people how to use it, run agencies, and organize local Clay Clubs. A lot of those people have played a big role in making Clay successful. So when Clay raised more money, they gave members of that community the opportunity to invest at the same price as their institutional investors. You're already raising money, and you've got all these people who've invested a ton of time into making your product successful. Why not give some of them the opportunity to become owners too? I think about things like this a lot when we're working on growth at Town. Most companies are doing the same kinds of things. Raising money, building communities, running events, etc. But sometimes you can do one of those things in a way that feels very specific to your company. I think Clay did a great job with this. What does that look like for Town? Wait and see!

Someone asked me what I'd do if I were in my mid-20s and wanted to optimize for learning. At 24, I was living in NYC, and honestly I wasn't really focused on learning. I was happy just hanging out with friends, going out once in a while, making some money consulting, racing bikes (badly), and enjoying a pretty wide range of things. I was not, in any sense of the word, successful. I realized pretty quickly that this was probably not the answer they were looking for. I did eventually figure out a few things in life, and I've learned a lot in the 20+ years since. I just don't think I was ever optimizing for learning because it felt then (as it does now) that life is long and there's a lot of it worth living and a lot of time for learning. What were you doing at 25?

Every company ends up using the same marketing channels. So you have to find ways to make them your own. In the last three or four years, for example, a bunch of tech and AI companies have grown a ton through influencers, and that wasn't really a thing in tech before. At least not to the extent it is today. But most of the time you're doing the same stuff everyone else is doing. Ads, billboards, community, marketing emails, etc. So then the question becomes how you do those things in a way that feels like your company. For us, we have these little characters called Townies. They have a pretty specific visual style, and people like them. So we can do some weird things with them in ads or have our marketing emails come from people's Townies instead of from me (well sometimes they come from me too). It's still an ad campaign. It's still a marketing email. But I think you find enough of these little things across what you're doing, and they start to add up. People get a feeling for your company that's a little different from everyone else's.