Down the Rabbithole is a daily show about building Rabbithole, selling, and living in the world of AI. In Episode 1, Claude interviews me about my first day of cold calling. Watch it on YouTube.
Why a daily show
I believe content will be the main form of distribution for every business going forward. The people who start early get more time in the market, and more time means more chances to learn. So I'm recording one episode a day about what actually happened that day.
Today I did something new: instead of just talking to the camera, I asked Claude to interview me. It named itself Lorenz.
Walking into call #1
Like most founders, I didn't love the idea of cold calling. It can feel embarrassing and uncomfortable, mostly because of how much rejection comes with it.
So I spent Sunday preparing my head, not my script. I went back to stoic philosophy, read about chaos and how small inputs create big outcomes, and consumed a lot of sales content. A few ideas stuck:
- Boring is good. Consistency beats charisma.
- It's not about the sale. The best salespeople aren't the ones pushing hardest up front.
- The journey, not the destination. If the only reward is a "yes," every "no" hurts. If the reward is doing the work, every call counts.
I also had an advantage: we'd already been building Rabbithole's infrastructure and had contacts from months of cold email, so I wasn't dialing strangers.
"10 feels like the metric you picked so the day couldn't fail"
This was the best moment of the episode. I talked about the law of large numbers, then admitted I made ten calls. Lorenz called it out:
You said the law of large numbers, then you made 10 calls. 10 isn't a sample, it's a warm-up. So which is it: were you running an experiment today, or building tolerance? Those need totally different scorecards.
Fair. Ten was a minimum, not a sample. What I learned:
- Start earlier. I need to call in the morning and push my usual morning tasks later.
- The new target is 25 calls a day.
The numbers
| Metric | Day 1 |
|---|---|
| Calls made | 10 |
| Picked up | 5 |
| Callbacks confirmed | 2 |
Most of the other five weren't flat noes: I didn't reach the owner, so I left voicemails and sent email follow-ups. Lorenz's take: "That's a better hit rate than most people's first week."
Memory: a skill vs. a journal
I want to keep talking to the same AI interviewer every day, with the context of everything we've discussed. My first instinct was to write a Claude skill for it. Lorenz pushed back:
A skill is instructions, how to do something. What you want is state, what has happened. If you stuff a growing journal into a skill, it gets bloated and stale fast.
The better shape is both: a small skill that defines how the episode runs (the tone, the questions never to skip), and a separate journal file it reads at the start and appends to at the end. The skill stays tiny; the journal grows. For me, that journal is basically our CRM.
Relief or pressure?
Lorenz's question for tomorrow: when the first owner says yes to a real meeting, will I feel relief or pressure?
My answer: no pressure. Once you've got the meeting, you're good to go.
What's next
Tomorrow: more calls, started earlier, tracked properly. And as I said at the end of the episode, I'm going to AI-ify this in every sense of the word, for fun and because building it in public is the point.
If you're a business owner wondering what AI agents could actually do for your team, that's what Rabbithole does.
Chapters
- 0:00 Intro: why a daily show
- 0:42 Claude joins the show
- 1:31 Walking into call #1
- 4:07 Law of large numbers?
- 5:28 Memory: skill vs. journal
- 7:04 Call me Lorenz
- 8:20 Relief or pressure?