In this episode:
- Peter's thesis for the AI era: build building blocks, not end solutions
- Memo.sh update: 43 waitlist signups without a launch tweet, and a new workspace instructions feature
- The value hacking phase (Mike Maples Jr.) and betting on yourself as customer zero
- Phone.inc's new infrastructure: deadline day, three smartphones, and a 50-bullet testing list
- Why the phone system should be primitives, and everything around it wide open
- The shorts factory experiment: 30-something shorts in an hour and a half
Peter's been heads down on memo.sh. The waitlist is at 43 people, and he hasn't even done the "hey, this is what I'm building" tweet yet (just leaving the link in regular tweets, which keeps converting one or two people at a time). He also added a follow-up form after signup with three questions, and almost everyone describes the same problem they hope Memo solves. That's a good sign. Next step is a demo video (the new Osmo Pocket 4 gets its first job), then inviting the first users. He's leaning towards charging them from day one, mostly as a filter.
He's thinking about all of this through Mike Maples Jr.'s value hacking framework: a startup is a group of people trying to productize a unique insight, and before you worry about growth, you nail the value. Peter's bet is that he's customer zero for Memo, and that today's AI power users look a lot like tomorrow's normal users.
That leads into the big idea of the episode: in the age of AI, you should probably be building building blocks, not end solutions. Everyone can now assemble their own workflows with AI, so nobody wants something too opinionated. They want great, slightly opinionated blocks they can put together themselves. Peter's newest Memo feature doubles down on this: per-workspace instructions for the agents connecting to it, which turns Memo into a flexible storage and search layer people can build things with that he hadn't imagined. Niklas pushes back a bit (he still believes in super opinionated products, and thinks the blocks will need to get more opinionated as non-builders arrive), but they mostly land in the same place.
Niklas is recording on the day of Phone.inc's internal infrastructure deadline, surrounded by three smartphones and a 50-bullet manual testing list (testing telephony is a project in itself). The new setup is built around conference calls instead of logical phone lines, which makes the whole call life cycle simpler and way more responsive. They're moving their own number over now and customers after the summer holiday. His conclusion fits the building blocks theme: the phone system itself should be a simple set of primitives, and everything around it (MCP servers, APIs, webhooks) should be wide open. One prospect on a recent call has written 450,000 lines of code with AI in six months and just wants his phone calls in that pipeline.
They also debrief the shorts factory experiment (30-something shorts in 90 minutes), and Lea reports that the vlogs are changing Phone.inc's hiring: candidates show up already knowing the company. Niklas's takeaway: when you judge content like this, count customers, hires, partners, and investors, not just views.
---
Still Early is sponsored by Phone.inc: a business phone number, welcome greeting, call routing, and opening hours, all from an app on your personal phone.
