Specialists that coordinate, then ship.
A small team of AI teammates helps run this lab. Each one has a narrow job — routing, GitOps, cluster ops, tickets, or research. They hand work across. A person still makes the calls that need a person.
The problem
One general assistant is a poor fit for a lab that already has repos, clusters, and a ticket trail. A chart change, a sync check, a project update, and a research note are different jobs. Mixing them in one thread drops context, and someone has to restate the operating model every time.
The model
Five standing roles. Sparky is chief of staff: incoming work gets a lane, the others stay aligned, and only a decision that needs a human gets escalated. The rest execute in their lane and report back. Code leaves as a pull request. The plan lives in GitHub Issues. The teammates ship on the same GitOps path as the rest of the lab.
Roles
Sparky
Chief of staff. Routes work, keeps alignment, escalates only decisions that need a human.
Dev Bot
Engineering and GitOps. Repos, charts, runners, and pull requests.
Ops
Cluster operations. DevOps and home-lab Kubernetes — sync, smoke, and careful changes.
PM
Project management. GitHub Issues are the system of record.
Research
Digs and write-ups when something needs investigating.
Why this shape
The lab already ships through Git, GitOps, and issues. The teammates use that path. A narrow role keeps a chart change in the repo lane and a cluster change in the ops lane. A research note stays a write-up until someone decides it is a deploy. One chief of staff means a person is asked when a decision is actually required. What ships is still a diff someone can read.
← Projects