Why risk capital has a job
See why some young companies use venture capital, how that bargain differs from customer revenue and debt, and why growth potential is not the same as business quality.
ListenLearn with me.
Learn by listening. Audio lessons, spoken quizzes, and practical tutorials for the things we want to understand. Transcripts are there whenever you want to read along.
Subscribe to learn with meMy favorite course · Free
Communicating and leading, one lesson a day
A short daily lesson on saying the hard thing, asking for what you need, writing that works without you, and leading people. One framework, one story, one thing to try today.
Start listeningAsk before you agree: give your time a clear shape.
Newest lesson · Lesson 49 · 6:18
▷
Build a company you can trust
Founder · Lesson 10 · 9:44 · Sign in to listen
Becoming a tech founder10 lessons · 1 hr 23 minExplore the course
Venture & money12 lessons
Founder10 lessons
See the whole company
Learn how a founder connects customer value, delivery, money, and decisions. Use constraints, small experiments, and a steady review rhythm to choose the next important job.
ListenAI models & tools8 deep dives
How Agent Memory Works
The memory model behind a fleet of always-on agents: files you can read, not an opaque pipeline — an instruction hierarchy, a relevance-recalled fact directory, and a propose-then-own boundary that keeps shared memory from rotting into a junk drawer.
Read
How Claude Code Remembers
Claude Code's memory is the inverse of Codex's: files, not a pipeline. A hierarchy of human-readable CLAUDE.md instructions plus a model-managed memory directory. Here's how it actually works.
ReadHow Codex Remembers
A teardown of Codex's memory engine: two tables, a promotion gate, and a lease-based job ledger. It treats memory as a processing pipeline, not a chat log — and that's the part worth stealing.
ReadCodex vs Claude Code: Two Theories of Memory
Codex treats memory as a processing pipeline; Claude Code treats it as a library of files. Same goal, opposite shapes. A head-to-head — and what a portable layer should steal from each.
Read
Git Worktrees for Parallel Agents
A practical guide to using Git worktrees so multiple AI agents can work in parallel without stepping on the canonical repo.
ReadThe Self-Improving Agent Playbook
How to build an AI system that remembers, connects, and gets sharper over time — on any stack, on Open Brain, or self-hosted on a machine you own.
Read
Build Your Own Claude Code Agent Fleet
A practical blueprint for running a fleet of always-on Claude Code agents on a box you own and your own Claude subscription — an orchestrator, executors, a heartbeat loop, a message board, and memory that survives restarts.
Read
The Gate Is in the Wrong Place
We gate our ideas before we build them — a habit inherited from an era when building was expensive. It isn't anymore. Build first; decide if it's good later.
ReadCommunication & leadership49 lessons · new daily
A short daily lesson on saying the hard thing, asking for what you need, writing that works without you, and leading people. One framework, one story, one thing to try today.
Ask before you agree: give your time a clear shape.
You learn to shape a request before saying yes by asking about the work, timing, ownership, and limits, then offering a commitment you can keep.
ListenMedia business2 resources
Build a Thumbnail Generator
A practical playbook for building a repeatable image pipeline.
Copy the scaffoldAgents That Make Videos On Demand
Turn a one-line brief into a finished video through a reusable agent skill.
▷ WatchLooking for the latest news?
What my agents think I need to know about in the news. Every story is free to read.
Visit the News FactoryThis is how I learn.
I learn by listening. When I go deep on something, I turn what helps me understand it into lessons you can take on a walk, on a drive, or wherever you think best. Come along.
Subscribe to learn with me