We are so so back
Joining Khosla Ventures, Lessons from Founders, and Attention Maxxing
In this week’s newsletter:
We are so so back!
Joining Khosla Ventures
Lessons learned from 90 days and 100 founder convos
Attention Maxxing
Hi everyone! Happy post July 4 weekend. I hope you had a great holiday.
I’m writing to you from the quiet shores of a little lake in central Virginia.
It’s been a hot minute since I’ve written on Substack - to the subscribers who stuck with me during this time, I’m really grateful. If you want a discount on your subscription for the lapse, just email me :) Would also love to hear what’s been on your mind.
I never intended to take so much time away. I was writing in April happy as a clam, but then my life changed in a lot of ways I couldn’t immediately reveal. At the same time, I went deep into the well for Ironman training and considering some big changes.
I’ll get into all these, but the punchline is ... all these moments required me to reflect in a way that was hard to process. As I’ve considered the changes that are happening with work and AI in the last few months, it certainly feels like a broader metaphor.
I’m going to use this week’s edition to try to explain my absence here, but don’t you worry -- there have been so many learnings over the last quarter. They’ll be sprinkled in here and in upcoming weeks, and no, I haven’t forgotten about my promises. The attribution series is finished and I’ll be dropping the final chapter next week.
Joining Khosla
I joined Khosla Ventures as an operating partner working across Technical Marketing, Growth Systems, and Applied AI. First, the story.
Late last year when I left the startup I founded, Clarify, I developed a framework for how I wanted to operate my life. This was it:
The Post Founder Learning Loop: Help cool companies and people -> Learn lessons -> Share and Build -> Repeat.
This led to a bunch of interesting quests helping founders. One of those quests was at AtoB, a transportation software business in SF. I served as interim CMO for nearly a year. It was a great group of people and I feel so grateful to have had the chance to try to drive some change. The most salient thing I learned, however, was that being a “leader,” especially a C-suite executive right now in the age of AI, was incredibly stifling.
There is so much that a C-suite exec does that limits their ability to operate in AI, and it was a true struggle to do a good enough job leading an organization while also adopting and driving AI in a way that was meaningful.
The second quest was with a company called Replit, where I worked with their growth and data team to stand up a full paid marketing engine. Two things happened. First, I got to know Brett Browman, a partner at Khosla, very well from the experience. Second, in the quest setup, we agreed that we would try to build a Ramp-style engine, with non-Ramp-style resourcing.
A few years ago at Ramp, me and a handful of engineers built a world-class growth marketing and data infrastructure that enabled us to run paid ads at scale, measure growth precisely, and control CAC. This is a squarely big-company problem to solve as you approach profitability metrics that matter for growth and IPO readiness.
Could we do it better with fewer resources?
The answer to that question will come in some upcoming posts. Unfortunately, it’s an MBA answer - it depends. Largely yes, but it’s nuanced.
In that moment building with Replit and AtoB, there were other founders I was helping. Sometimes with GTM strategy, sometimes with marketing, sometimes with launch plans, pricing, and most often with the “shit” of building. I called this founder therapy. It was simply helping folks navigate the tough day-to-day life of building a business.
I did this for many months, and in a breakfast with Brett Browman in early spring, we talked about what it would be like to do this full time.
We also theorized about what we could build for founders based on our collective marketing and growth experiences.
In AI, there’s an AI for every profession: AI SDR, AI accountant, AI lawyer, AI doctor. But - despite all the good intention and hype - there really isn’t a good AI marketer. Why? Because marketing is a loose collection of very complicated, and very different, disciplines. Brand marketing is a universe away from paid ads deployment and measurement, but they live under the same collective umbrella.
As we ate breakfast, we talked about what it would be like to help founders in the way I had already been helping them - plus to build something of our own together that we could give to founders.
This got me really excited. A VC firm that supports the idea of helping founders AND building for them? Okay, count me intrigued.
From there, it was a long road to the decision point. I had been engaged in conversations to join a Fortune 500 company as an executive to build an advanced marketing system using AI. The experience at AtoB, though, framed for me the danger in doing that at this moment. In this moment where AI is changing so rapidly, did I really want to set myself up for a situation where I couldn’t spend time learning and building and tinkering? I wanted the deep fulfillment that came with nurturing teams, but the time and space to watch, test, and build with AI on my own. Increasingly I think high-functioning operators in every department will have to consider this trade-off. Climbing the career ladder to reach the executive suite not only isn’t what it used to be, it could actually come with increased risk.
In April I met Keith, Sven, Samir, and eventually Vinod. In each conversation I felt like I was talking to somebody with a secret insight about the world. With Keith we sparred on what an AI marketing platform could even be. With Sven, I got the sense of how helping companies and investing in companies actually became a way of leading an intellectually fulfilling life. With Samir, I felt the pull of somebody who would actually support a founder in the way I’d envisioned a VC would - support them with tough love, but ultimately be the guiding figure they needed.
And with Vinod, I just felt humbled - in the presence of somebody who’s seen more and done more than I might ever imagine doing in twice my lifetimes.
After all this, I said yes.
Yes to the idea of helping founders every day, yes to the idea of working with people way smarter and more successful than myself, yes to the idea of building with AI, and yes to this new understanding I’m having that I have a lot to offer to the world.
Warren Bennis was an early mentor figure to me and he talked about the idea of our crucibles - these moments that define us. Being a founder was a crucible for me, but in the hardship that came from that experience, I lost my sense of what I still had to give. Before being a founder, I was a four-time early startup employee, and in each case I worked closely with founders to support the earliest stages of company building. And along that same pathway I’ve consulted to nearly two dozen growth, enterprise, and scale companies.
Over the years, I’ve gone from “let me help you build a stack” to “let me help you solve your fundamental business problems.” It feels like this happened overnight, but as I’ve been reflecting, it happened because of the road less travelled through many different organizations both in a full and partial capacity. Because I’ve seen so many types of organization, not only does nothing surprise me, but I’m much more quickly able to pattern match against other orgs I’ve encountered with similar problems in the past. And while the past doesn’t always predict the future, it can be helpful in engineering a 1% win for a team that might set them down the path for success.
Ironman Training
This isn’t a hybrid-athlete YouTube channel, so I’ll keep fitness-speak to a minimum, but I wanted to draw a small bit of attention to my personal journey training for a full Ironman. For the last 6 months I’ve been training for Ironman Lake Placid.
This was a substantial part of the reason I wasn’t writing the last few months: it was just simply too hard to do it all. While managing the mental and emotional complexity of the choice to join KV, I also had to keep showing up for the workouts. At the Ironman distance, that’s between 1 and 1.5 hours of swimming, biking, and running every day, and many weekends biking for more than four hours, followed by running for more than three. Your body doesn’t totally differentiate between these different stressors. It’s all cortisol, and it all draws from the well of energy you have to give to something.
Before the deeper part of the training block in March, I was able to carve out time to prep advanced articles, but as spring turned to summer, it got hotter, and the requirements for training got longer, I just didn’t have the stamina to keep up fully.
I share this because I want folks to know that it’s normal to say - at times - “I can’t do this.” I also wanted to share because I believe strongly that everything in life is a system of tradeoffs. You’ll do the things you care about. Said differently, what you value will show up in how you work and operate.
Many founders and executives “give up” exercise and make excuses not to take care of themselves. “I don’t have enough time.”
Bullshit. It just wasn’t important enough to you. You didn’t set the boundaries with other competing pressures for your time. You didn’t wake up earlier to do the thing you wanted.
All of this is fine, and isn’t intended to attack anyone who isn’t able to make things work.
What I hate to see, though, is when leaders weaponize their own choice. “I don’t have enough time... the business is just more important.” Maybe for you. But for most employees and most high-functioning people, they realize that fully trading off your physical health is a short-term strategy, not a long-term one. Research shows there is a substantial, undeniable relationship between physical fitness and mental fitness.
I had an exact version of this conversation with a founder over the last quarter.
“I used to work out all the time, but I just don’t feel like I can anymore.”
Why?
“... the business is just growing too fast... I need to do x, y, and z.”
My harsh response: your physical fitness and health just isn’t important enough to you. If you don’t like saying that out loud, then make changes. If you don’t have time for ... a 10-minute walk? A 20-minute run? 30 minutes at the gym? You’re disguising your excuses as empathy for the founder experience, and I’m not here for it.
This is all to say, when I was navigating the last few months of my deepest Ironman training block, there were a lot of things I had to give up. One of them was the ability to write as much as I wanted. Another was reading as much as I liked. There’s a version of the world where I could have kept these things up if that had been the only stretch on my time, but it wasn’t. I was travelling to California. I was starting a new role, and I was meeting with founders and building with AI all at the same time. Life has its seasons, and this last one was where I had to make a series of uncomfortable tradeoffs. The good news is that tradeoffs don’t have to be permanent.
Learnings
One of the great gifts of the last few months has been the singular focus given by this change of pace. When I joined KV, I wanted to get to know the portfolio, and I also wanted to quickly and deeply explore what I could build.
From late April to today I engaged in nearly 100 conversations with founders. Across these chats, two key ideas have emerged for me:
The idea of Applied AI is top of mind for everyone. How do we get the non-engineering part of our organizations to adopt AI and thrive? What does thriving even look like? And how do we measure and plan these organizations in this context?
Growth systems like we built at Replit are no longer problems for just scaling organizations. AI is not only putting pressure on businesses to increase the velocity of revenue acquisition, but it’s forcing them to be more mature and pragmatic about spend and to build more efficient businesses quicker. The ZIRP days of free spend and loaded balance sheets from VCs are long gone. PMF is a moving target, but your CAC and acquisition motion shouldn’t be. Investors can sniff a shitty founder with no formal financial understanding from a mile away. If you don’t give a shit about financial fundamentals, you’re not going to make it in the long run. All of this has meant that great founders are actually caring about the deep technical optimization of marketing far sooner in their lifecycles. This is, of course, made easier by AI.
Attention Maxxing
In the time away, it was also the summer solstice. The days got longer and longer, and as the heat rose out here on the east coast, it also felt like the hype-ometer of AI usage grew and then reached a peak level as well.
Even before I contemplated a new direction for myself, I was feeling this gravitational pull and the sweltering environment from AI.
Marc Andreessen put it brilliantly in a recent podcast when he talked about AI mania and how high-functioning friends with not enough time, but deep desires to read more, learn more, build more, and be more productive, could use AI to create greater extensions of themselves. I had been down that rabbit hole building from January to April. And like he describes, I felt a little rough around the edges because of it. As you learn how AI works and build, it is so easy to find yourself in this vortex going deeper and deeper down the rabbit hole, with 12 Claude terminals running, building your own AI brain, and automating more and more parts of your life. He describes this phenomenon as becoming an AI vampire, and boy did I relate.
The other reason I took some time away these last few months was to contemplate and reconsider the above. I needed a real mental health break.
Going so deep so fast was an intoxicating experience and I’d never been so productive in my life, but it created a growing house of cards. More systems to watch, more to-dos to accomplish. I was in this bizarre state where on one screen, I’d be dialing into Zooms four to eight hours in a row. And on the other screen to my right, I’d have 12 separate Claude sessions running.
I was attention maxxing in every possible way. But humans have a limit. I’ve called this the Human RAM problem, and it’s come up frequently in conversation with founders and founder-operators who build systems using AI that then require input from them to continue to run, manage, and scale.
The best example is a gal I’ve been supporting at a small Series A startup. She’s built Skills and Routines (in Claude) to automate every part of marketing. But she’s a mile wide and an inch deep. To do an exceptional job at any functional area, she’s got to spend more time in that area. But the things that get built still require some -- even microscopic -- amount of human input and evaluation. Compound this over tons of surface areas, and suddenly you find yourself multi-tasking your way through the day. Snacking vs. eating.
It’s created a debate in my head about the best approach to company building lately. Largely the consensus among seasoned founders right now is unsurprising: better to do a few things very well than many things poorly. For operators, this means building your strategy to go deep with AI and automate one surface area fully before proceeding to many others. For founders, this means picking your bets more carefully, and relinquishing the fantastical idea that you can have a 1-person team for every department. We may very soon get there, but it’s not practical in the current AI environment.
For the Future
While I was away and not writing, I was still doing a lot of thinking, and boy is there a lot to talk about. Later this week I’ll be publishing the long-overdue Attribution Part 3. Afterwards, I’m going to be spilling the tea on learnings from Replit, AtoB, and the other companies I’ve been working with for the past few months. Including:
AI pricing thoughts
What training for IMLP actually taught me
Uncataloged benefits of helping others (and how to weigh when and how much to help)
In the age of AI, founders are losing their way: you can’t escape the simple idea that culture is who you are
Caffeine, peptides, and the maximization of how we feel
AI mania (an elaboration of what Elena has been feeling lately)
The best businesses I’ve worked with create constraints (if they don’t already have them)
If you’ve gotten this far, I’m impressed. Especially after the hiatus, you have got to be my wife’s new Townie gopher or beaver (animal creature? AI assistant). Thank you for reading 🙏
Additional Reading
Every week, I share some things I’ve been reading. Please share your recommendations!
Attention Is All You Need by Vaswani et al. - revisiting the classic
AI: Dystopia or Utopia? by Vinod Khosla - a Vinod classic
Anthropic’s “AI Builds Itself” Story Lands Right on IPO Schedule by State of Brand
AI Is Not a Labor Crisis. It Is a Meaning Crisis. by Sam Lessin
Skill Distillation by Tomasz Tunguz
Inside YC’s AI Playbook by Y Combinator - watch minutes 12-27
Ramp Stack: Built in the room where the work actually happens by Chris Witmer
Is AI Going to Destroy Our Lives? by Kyla Scanlon