Getting started
AI transformation
Bottom up means giving individual people AI harnesses and skills. Top down means one company backend that runs work in the background. Both matter, and only one of them is getting attention.
There are two directions AI enters a company, and almost all of the attention sits on one of them. This page is about the other one.
Bottom up
Personal AI
One person gets faster with an AI harness, and the practice spreads to the team. It only runs while you are at the keyboard.
Top down
Company AI
The company runs one backend, and every person and team connects to it. It runs without anyone present.
Bottom up
Bottom up is personal AI. You give individual people coding harnesses, agents, and skills, and each person gets more done. It starts with one person, spreads to their team, and eventually changes how the company works. Dave uses this heavily inside Datalumina and it works.
It has one limit. Someone has to be there, in front of the computer, steering. Close the laptop and the productivity stops.
Top down
Top down is company AI. One shared backend runs the work, and people, teams, and other systems connect to it. Work arrives from a webhook or a schedule, a worker picks it up, and the result lands in the systems people already use. Nobody has to be present. This is the half almost nobody is talking about, and it is the kind of system this series builds.
The top-down system contains very little AI at runtime. Most of it is deterministic backend code, on purpose, because reliability is the entire value of something that runs at 02:00 with nobody watching.
Why you need both
The two directions are not in competition, and a company that only does one misses half the gain. Personal AI raises the ceiling on what your people can do in a day. Company AI takes the work that should not need a person at all and moves it into the background permanently. Personal AI works with you, and company AI keeps working without you.
The plumbing tax
Replacing a five-minute Zapier scenario with your own backend costs a server, a deployment, auth, and a webhook endpoint. You pay it once, and then a new workflow is close to one prompt.
The architecture
How an event-driven automation backend fits together, from Cloudflare and Caddy at the edge to Celery workers, the event ledger, and the managed services behind them.