I build and grow companies using AI.Work with us →
How we run

This page shows the real thing: the fleet, the services, and what a person keeps. The map is on How we think. The machinery is on Earthworks. Recent cases are on Examples.

Plate II: the operator's desk as a specimen arrangement, living agent-forms laid out around a working desk, the human hand at the center, modelled in layered aquatint wash

How we run

the setup

Set Earth keeps a fleet running across its operations.

The goal is to never be the bottleneck, so the work runs on fleets of agents. An agent is a software assistant set up to do real work on its own: it reads files, runs commands, writes drafts and code, and checks its own work before showing a person. A fleet is a team of them assigned to one job under a person’s direction. A typical week has somewhere between 10+ fleets running 2-50 agents each. Some fleets are working on and for companies, some on research, some on a build, a review, a memo, or a problem still being thought through. Some are working; some are waiting.

To make this work, each agent gets a place to work, the memory it needs, and its list of work. That context is part of the agent. A fleet might sit idle for a day; the context is still there when the work resumes.

The tool at the center is called Herdr, with Pi underneath as the harness. Everything stays visible in it: the agents, the workspaces, what is blocked or moving. Agents have different roles, and the work picks the model it needs: cheap and fast for the routine, careful and expensive for the judgment. The newest development is a “Collective” agent framework: a fleet operating on its own, relatively unsupervised. Wild times.

Behind the workspaces are the services that keep things working: where the memory lives, where the work is tracked, what runs on a schedule, and the checks that notice when something breaks. Some work starts when a person assigns it. Some starts when a timer, event, or service says it is time. With the new Collective, the fleet can also start work on its own; what it starts still comes to a person for review.

Herdr, in use. The workspace around the fleet: separate spaces, live agents, and the work a person is actually steering.

the why

The expensive and exhausting part of running operations like this is the judgment.

More work used to mean more hours, or more people, or both. This way, the expensive hours go to the framing, the thinking, and the decisions. The cheap hours run as standing work in the background. That gives a person much more scale than they used to have.

The fleet also changes what can be taken on at once. The old way was one thing at a time, with everything interrupted when something urgent arrived. Now several operations stay warm at once.

the operator's day · loggedfig. 2 · one tuesday
38 fleet decisions, logged
5 operator moments
2 lines edited by hand
1 precedent set
06:58 [auto] overnight runs complete — 3 fleets · 11 workspaces · 2 deploys shipped
07:12 [auto] workboard digest ready — 38 items cleared, 2 flagged for you
07:14 [you] “hold the renewal review until the sponsor call” — routed, logged
09:30 [auto] board update drafted — context: chair reads numbers first · risks drafted page one
09:34 [you] edited two lines — “risks on page one, not page four” — held
09:36 [you] send — the things that matter, a person edits and sends
11:02 [auto] invoice chase · reminder drafted · step-2 rule · 9 clean prior
11:02 [auto] send (routine: no edit needed · precedent held since August)
13:15 [auto] panel returned — 5 seats reviewed the draft · 2 fixes applied
13:41 [you] ruling — “fix then ship” — logged as precedent
16:20 [auto] month-end close — bank rec chased · accruals queued behind it
16:44 [auto] site deploy checklist green · build verified · ready
16:45 [you] “go ahead and deploy” — logged
17:20 [auto] outcomes recorded — tomorrow’s work starts from today’s log
decisions logged · nothing collapses off the record 17:20
Fig. 2 — one Tuesday, as the fleet logs it. [auto] rows are the fleet deciding with the context it holds. [you] rows are the operator: short, decisive, logged. Every ruling is logged, and the next run of the same work starts from it. A person always edits and sends the things that matter. OCT 2026

the handoff

The context survives the prompt.

A task can begin in a repository, pick up decisions from memory, use a service, and leave a result in a task graph or review queue, or just in a document.

A person sets the context, assigns or redirects the work, and inspects the results. To be fair, there is a lot of steering along the way.

the collective · flowingfig. 3 · 00 events
the record · 0 board events · newest last
tskindseatsubject
from the board log · nothing invented
Fig. 3 — the collective, flowing. No board events in the record yet; the lanes and the table below are the empty log. OCT 2026

what a person keeps

The decisions, the framing, and the boundaries stay with a person.

A person decides what work gets done, whether a result is good enough to use, and whether a system keeps running.

No autonomous inbox replies here. Tried it; not good. AI generates many drafts from materials, but a person always edits and sends.

The agents also produce too much stuff: drafts, artifacts, digressions. It is a lot to wade through. There is intuition in what to pay attention to.

The system also stays separated by company and purpose. Memory keeps context between sessions. It is not one giant pool. Operational state keeps the current facts, where they came from, how fresh they are, and what changes are proposed.

the cost

The system runs on maintenance.

It’s nowhere near perfect. Memory drifts, agents need their boundaries reset, and a service update breaks a connection. A task graph needs repair, a timed job fails. Nothing breaks silently, though: the agents report their own failures, so a person sees what broke and decides the fix. The system has to be watched, updated, and occasionally simplified.

the difference

How this is different from the old way.

Ten years ago, this was one company, a handful of direct reports, and a calendar. Work moved serially: a person in every meeting, on every thread, and often the bottleneck.

Now several operations stay warm at once: a company, a build, research, a talk. Memory keeps the context between sessions, and a new standing workspace with the right role and review loop takes an afternoon to spin up. Braydon still makes the judgment calls, now at review checkpoints instead of at the front of every task.

The fleet

An agent is a workspace with context and a job.

Fig. 4 — the operating loop. The agents keep work moving through the system. We set the context, redirect the work, and decide what to trust and keep for the next state. OCT 2026

up to speed

What it looks like to take on this way of working.

Getting up to speed starts small: one build, not the whole system. The first month is the build: what gets made, what a person keeps. After that it runs as standing work the agents check on a rhythm. The decisions get ninety minutes a week, and the numbers get ninety minutes a month.

Nobody trains on the tools. The way of working is learned from inside it, the same way it was built. By month nine your team runs more of it and Set Earth reviews.

The terms and lengths are on Work with Set Earth, fit call included.

the outcome

The system earns its keep.

Six companies and Set Earth itself now run this way, and a week that used to end with everything unfinished now ends with most things in a decent shape. The system itself is infrastructure, and it takes watching and repair. That trade is worth it.

NextSee how the work is shaped →


Set Earth Inc · Contracts and invoices under this entity · Reply to b@mcco.us