Dad, You Actually Wrote Code?
I’ve been heads-down the past few weeks building stuff with AI tools and… wow. Multiple takeaways, but the first layer would be the vibe coding stack is already stupidly powerful. If you have even a baby-level grasp of how apps fit together, you can go end-to-end and ship a mini-app. Run a few loops, standardize your favorite backend skeleton, and you can mass-produce. In 2025, going from idea to MVP is literally one step. Blink and you’re deploying.
AIBJ#
Bill Hsu (friend, co-conspirator, co-worker) and I started asking a simple question: as VCs, where can automation and AI give us 10x output, or let us be 10x lazier without anyone noticing. We call the project AIBJ.
Yes, it’s “AI-Bill-Johnny.” No, that’s not the only meaning. If you guess the others, DM me and I’ll buy you coffee.
We’re tracking a bunch of projects in Notion, each meant to snap into the VC workflow:
Branding → Deal sourcing → Diligence → Execution → Post-investment tracking
At every step we’re asking: what’s the smallest thing we can automate that creates the biggest dent. We pick either obvious low-hanging fruit or projects with comically high ROI.
A few examples:
- I just shipped an agent that watches a curated set of X accounts (Paradigm, a16z, etc.) and flags their new followings. An LLM then classifies: is this a project, what stage, and is the product our cold-reach persona or a waste of time. It’s like having a paranoid intern who never sleeps and doesn’t ask for carry.
- Bill built an agent that slurps DocSend decks from inbound, drops it into LLM with Web Search, and spits out a pre-meeting cheat sheet.
We’ve got a few other agents in the pipeline, too. Like one that automatically scrapes our portfolio’s hottest tweets, then auto-writes and fires off quote tweets to help us piggyback off their buzz. Another one is designed to automatically surface top-performing DeFi protocols based on on-chain data.
A Catch-up Call#
It was also on this morning that I hopped on with a friend at a tier-1 crypto VC in the US. We compared automation notes and he pulled me into a small group chat of ~10 folks across web3 (mostly on the investing side) who are quietly building internal pipelines. The consensus was obvious: VC is a legacy industry dressed in a Patagonia vest. There’s a ridiculous amount to automate. Knowing I’m not the only goblin wiring cron jobs into dealflow is comforting. Put yourself out there and luck compounds.
Moat#
As investors we obsess over moats. If a hungry competitor shows up, what keeps you from getting eaten. The question snuck up on me this year: in an AI world, what’s my moat. What’s humanity’s moat.
Generated by Gemini. Yeah maybe AI’s not quite there yet..
The more I use these tools, the more real the fear gets. In college I ping-ponged between Finance and CS/DS. I joined a startup and tried to split the difference, then wandered back toward finance via VC.
Let’s oversimplify to make the point: Finance vs CS.
For a decade, software ate comp packages. Big tech threw six to seven figures at perfectly fine engineers. A junior could pull $150–$200K and start a YouTube channel for dessert. This year, junior SWE roles got slashed and a lot of that looks irreversible. The “major in CS, profit forever” narrative is wobbling.
Finance isn’t exactly a safe haven either. Quants are feasting on AI, and “business analytics” grads keep popping out of every oven. Unless you’ve got closed data or a structural angle, what’s left that a garden-variety model can’t chew through.
It feels like a paradigm shift mid-render. The meta is unsettled. Maybe there won’t be another monolithic “golden era job” again. So what actually differentiates a person now?
Again, I have a mid-tier IQ and am wrong most of the time. But if I had to bet on traits, I’d pick these.
The Shortlist#
Voracious learner
I genuinely think passion is still the cheat code. Most of us sit somewhere on the same bell curve of “natural ability,” so the separator isn’t IQ, it’s who’s willing to go full goblin mode for months reading, building, breaking, repeating, long after everyone else taps out. If you actually care, you’ll keep showing up when it’s boring, and that’s where compounding kicks in.
Be like water
When I say “be like water,” I mean the ability to unlearn. Drop the ego, abandon your old playbook, and pick up the tool or framework that’s objectively better, even if it nukes half your prior beliefs. That’s hard because it feels like admitting your past self was wrong. Do it anyway. In this era, stubbornness is expensive; flexibility pays.
BD/Sales muscles
No matter how fast products evolve, someone still has to move the thing. The job is simple but not easy: pick a target persona, test angles, get binary signals (yes or no), and then iterate your BD model like you would a product. Messaging, channel, offer, follow-ups: measure it, tweak it, ship it again.
Aggregator type of people
Think “Jupiter, but for humans.” It looks like BD, but the core skill is reading incentives across founders, users, investors, partners, maybe even regulators, then routing time/capital/attention so a deal clears that none of them could pull off alone. It’s multi-dimensional matchmaking: translate needs, align timing, compress trust gaps, and get everyone to yes.
Attention is all you need
AI is freeing up a ton of human productivity, which, let’s be honest, mostly flows into entertainment. Everything is fighting for your eyeballs: PS5, Netflix, podcasts, YouTube, whatever. Where attention goes, monetization follows. So write more, post on Threads, make videos, experiment in public. It might feel cringe on day one, but the EV is positive and the compounding is real.