Writing to you live from San Francisco!!!! (and also) two glasses of Sauvignon Blanc deep, a little sleep-deprived, and very aware that it’s only Wednesday (well Thursday, by the time you open this)

This week has been one of those rare weeks that stretches you in all the right ways.

I’ve had dinners, meetings, nonstop conversations, and one very clear realization: I don’t hate my little home office. I needed contrast. I needed new rooms, new energy, and new people to bounce against.

And I needed to be reminded of this: I’m still the person who gets the biggest dopamine hit from results.

I love spotting a human problem and turning it into something useful with tech. But lately that process has felt clunky, and if you’re in a corporate role trying to build AI tools inside enterprise constraints, you probably know exactly what I mean.

I work full-time at a venture capital firm. My plate is full: board travel, comms, diligence, data rooms, and all the invisible operational glue that keeps things moving. I do not have founder-level time to go deep on one thread for 12 hours a day (but do any of us really?!)

And yet, I still keep opening my terminal every week.

Even when it takes 17+ clicks to get where I need to go.
Even when security and infrastructure make simple things beyond smooth.
Even when it feels like everyone else showed up to the race on motorcycles and I’m rolling up on a bicycle.

This week, we flew in a friend I hadn’t seen in years who now advises on AI in venture settings. He led a nearly five-hour session for our team. It was sharp, practical, and honestly humbling in the best way.

Afterward, I pulled him aside and showed him my setup.

His feedback was kind and direct: you’re doing the best you can with what you have, but your system is fighting you. This confirmed something I care deeply about saying out loud:

There are a lot of people building AI for startups.
There are fewer people speaking to the operator inside a corporate job, with real constraints, who still wants to build anyway.

That is who I write for. That is who I am.

I also want to share two ideas that helped me this week and might help you too.

First: I think loss aversion explains a lot of why smart people stay stuck: it’s real, but it’s not fixed.

Loss aversion is the tendency to feel potential losses more strongly than equivalent gains, and it can quietly shape our decisions more than we realize. A 2022 Theory and Decision paper found strong evidence for loss aversion in a large non-student sample, while also showing meaningful variation across people and decision types. A 2020 Psychonomic Bulletin & Review paper found that when researchers changed the decision environment, people’s gain/loss sensitivity shifted substantially, and in some cases even flipped.

Translation in normal-person language: sometimes what feels like “I’m not brave enough” is actually “this environment is training me to avoid downside.”

That matters for AI adoption at work.

If your workflow is high-friction, high-judgment, and unclear, your brain will protect you by overvaluing safety. That is not weakness. That is design.

Second: agency is often borrowed before it’s built.

Scott Galloway talks about greatness as the agency of others. I felt that this week in real time. We could have stayed in our own loop and stitched together articles and podcasts. Instead, we invested in someone who could see around corners faster than we could alone.

That is not dependency. That is leverage.

If you’re waiting for permission to invest in your own learning, this is your sign.

Not performative learning.
Applied learning.
Learning that buys your time back and expands what you can own.

I saw this in a very practical way today when I spent 36 minutes booking four flights for my boss. That’s not beneath me, and it still matters, but it’s also exactly the kind of repeatable workflow AI should help absorb. So I started building a better process: structured intake, cleaner prompt scaffolding, and a draft approval path that reduces back-and-forth. It’s not full automation yet (enterprise reality), but it’s already moving from scramble to system.

And that is the whole game right now.

AI won’t just replace jobs. It will absorb tasks.
Our advantage is how intentionally we reinvest the time it gives back.

If you want a practical starting point, try this this week:

Curiosity is still the luxury.
Not because it’s cute.
Because it compounds.

More soon. I’m still digesting the infrastructure changes I learned this week and I’ll share what actually works, what breaks, and what’s worth your time if you’re building inside enterprise constraints.