There are two kinds of pivots and they are very different

The obvious AI bubble, the Hamlet seed round, etiquette school, and more

---

This week was SF for our small holiday party + some meetings + the etiquette school book launch (more below).

I’m getting wrist surgery next week which will make typing hard. Expect shorter prose.

---

Pivots: changing the vector or the destination

Pivots can really suck for founders and we don’t want to pressure people to “make it work” when it’s not working/isn’t gonna work. In general, we don’t support people trying stuff just to try it when the original idea has clearly failed

But there’s a lot of space between that and the obvious fact that in very early companies you’re largely betting on people and general directions; the rest is subject to change and experimentation.

So what kinds of pivots should you do/do I (generally) support? I think there’s basically two kinds:

1. Type 1 pivot: change the angle of approach on a consistent vision or focal point. This is just wiggly early stage experimentation and iteration with consistent commitment to a vision of the future.

2. Type 2 pivot: change the end point and thus the way you get there. This constitutes changing the company entirely. It almost never works because you’re starting over at a disadvantage (less money and time).

Much of my investing/process with founders is about working backward from big themes and stories to companies that are worth owning in that future and finally how to start working toward them today. Within that framework, pivoting/changing the angle of approach is a question of “how” not “what” or “why.” So type 1 pivots are not only common but good.

And anecdotally, when founders ask me for an opinion, their pivots usually fail. But when founders tell me about a decision, their pivots usually work/go well.

Make of it what you will but I think it largely comes down to conviction and speed of execution. This is over enough examples and time to be a pattern.

Of Course We’re in an AI Bubble

A couple weeks ago Fred Wilson referenced Evan O’Donnell’s work tracking token usage to demonstrate that we’re not in an AI bubble. He concludes that “The current infrastructure spend rates are justified if the current rate of AI usage continues. If the growth rates start to decline, there could be trouble.”

But this is the wrong mental model/wrong question to ask.

In the 2000 crash, (revenue) multiples were blown out. And again in the 2021 private tech bubble. But here there is no doubt about the (token/revenue) growth being real. The multiples, at least in the public markets, are sane. Instead, this time around the revenue/usage itself is dubious.

No sane person is arguing that AI companies aren’t growing exceptionally quickly: from VC backed apps to Nvidia this is obviously true. The questions are in the retentive quality and earning power of that growth. Valuations, obviously, are forward looking.

In more strictly market terms, Felix Salmon claimed that we’re not in an AI bubble because 1) it’s not a get rich quick scheme 2) there are no greater fools 3) it’s not speculative. This is wrong on all three counts.

1. There are get rich quick schemes. There are ample ways to get rich very quickly from AI investing and many people are cashing in on. These include: selling stock in the hot companies (there are active secondary markets and they function like public companies), raising capital as an SPV provider and charging origination fees (the practice has been widely documented), and most significantly just raising huge sums of capital for funds (mgt fees to invest 500M at a time will make you very rich upon signing the docs).

2. There are greater fools. It’s not strictly speaking “retail” but there is absolutely a retail-like cohort of HNWI clamoring for exposure here. They are/will be the bag holders, as very likely will Joe Public both through the IPO market and government guarantees.

3. It is highly speculative. Just because the speculation is (mostly) happening in private/semi-private markets, doesn’t mean it’s not speculation. With #1 and #2, there are ample opportunities for purely speculative investing.

You don’t need to work my word for this - the inflaters of the bubble will say so themselves! The argument is not that AI is not a bubble but rather whether or not bubbles are good or “worth it.” It’s a Carlotta-Perez-style argument. TLDR bubbles are good/have utility because they lay the groundwork for R&D and buildout of ultimately useful infrastructure.

The question isn’t “are we in an AI bubble.” We are. The question is “so what?” What do you/we do or do differently?

Hamlet’s seed round

Most of the power that matters to businesses building in meatspace lives in four hour council meetings. And most companies only experience local politics as a blunt yes or no on a project. The actual logic sits in badly indexed hearings, PDFs, and livestreams that nobody watches. The arcane obscurity of local council meetings is one of the key mechanisms by which NIMBYism holds back dynamism and abundance.

Hamlet is trying to flip that. Take the raw exhaust of local democracy and turn it into something you can actually reason over when you decide where and how to build. Hamlet ingests agendas, minutes, and meeting video across cities and turns them into structured intelligence. Covid put local government online and LLMs made their proceedings parsable.

Customers developing data centers, energy projects, and housing use Hamlet to see when issues, parcels, or asset types they care about show up on an agenda. They can search across transcripts, pull exact quotes, and understand how individual councilmembers think about a given use. Hamlet helps critical industries build better, faster, and with more community engagement/support.

Over time, Hamlet is building a local decision graph and operating system for doing business in cities: agents and tools on top that handle site selection, risk scoring, and briefing prep. It also helps unblock a major artery of permitting, community engagement, and local development. If you care about an abundant future in the real world, you have to care about councils, boards, and the information systems that sit on top of them.

We are excited to share that we have led Hamlet’s seed round, with participation from Animo and Crosslink. Hamlet has raised nearly $10M to date.

Hamlet CEO Sunil Rajaraman is the most Abundo-pilled founder we have backed to date. Sunil is a repeat founder and seasoned operator, having founded or helped build Scripted, GoodRx, and Metromile. Most importantly, he has real vision and conviction for making local government a more effective, operable interface for operators and communities.

You can read more about the company in Techcrunch and you can watch Hamlet TV on Apple TV and Youtube. Hamlet TV is hilarious - it’s just the most out of pocket moments from local council meetings as a compilation.

Links

• How this 21-year-old college student used AI to build ‘Learning with Lyrics’. This is my favorite instagram account right now.

• You’re on Ozempic? How Quaint. We’re all gonna be on GLP-1s and they’re gonna be better and better (they might not even technically be GLP-1s pretty soon).

• Democrats have their own extremist problem.

Etiquette Guide

Pictured: Slow founder Oren Falkowitz reading the guide....

Slow’s Etiquette Guide for modern business has launched on Amazon! Tech and startups aren’t outsiders anymore. We’ve won.

Slow is doing important work to make sure people show up accordingly with poise and grace suitable to the task of running the world. The titans of industry used to build museums and cultural institutions. Etiquette school might be the start of reviving that noblesse oblige among founders.