@readmodelmag
MODEL
Subscribe

AI News

The CEO of Coinbase now has a chief of staff. It's an agent.

SummaryOn Nov 7, Coinbase CEO Brian Armstrong described his personal work method: a chief of staff agent overseeing lead agents, all feeding one local priority queue he clears mostly by voice, not a Coinbase-wide rollout.

· MODEL Editors · 1 min read

Some links may be affiliate links. Disclosure

MODEL cartoon on a cream background. Black header reads: The CEO hired a chief of staff. It's an agent. Left: an org chart of cute generic box-head robots. Top purple robot labeled CHIEF OF STAFF AGENT, oversees the leads. Three lead-agent boxes below: LEAD AGENT Area A (teal), Area B (blue), and Area C (orange). Four sub-agent boxes at the bottom: write code (green), test it (yellow), chase reviews (red), coordinate (gray). Caption under the chart: One queue instead of dozens of chats. Right: a PRIORITY QUEUE panel labeled simple local web app, with two SCREENSHOT product-review cards and a VIDEO DEMO of the feature card. Below the cards, a generic human silhouette wearing a headset (no face, no likeness) next to a green APPROVED badge, captioned clears the queue by voice. Footer note: Illustration. Generic robots, no real faces, no logos. Watermark MODEL · readmodelmag.com.
A chief of staff agent, lead agents, and one local priority queue. Armstrong describing his personal setup on X, Nov 7, 2026. Illustration by MODEL. Illustration by MODEL. Original. Image source.

What he posted. On Nov 7, 2026 at 8:59 AM PT, Coinbase CEO Brian Armstrong (@brian_armstrong (opens in a new tab)) wrote on X that he had "changed how I work over the last few weeks." This is Armstrong describing his own method — not a Coinbase product launch, and not a claim that Coinbase employees all work this way.

The org chart. He now has one lead agent for each area of focus, running all the time, and a chief of staff agent that oversees them. Each lead manages its own sub-agents that write code, test it, chase reviews, and coordinate with team members. He did not say how many agents he runs, which models or tools power them, or which areas the leads cover.

The queue. Everything he needs to decide or review comes into one prioritized queue, "a simple local web app." Each product review includes screenshots and a video demo of the feature. He clears the queue in batches, mostly by voice and keyboard shortcuts. The leads also bring back ideas he hadn't considered.

"Brian brain." Working from one queue, instead of switching between dozens of chats, "has cut a lot of mental overhead," he wrote. The system keeps learning his preferences and distilling his judgment into a "Brian brain," so its recommendations and first drafts get better before they reach him. The goal is fewer corrections per change. He did not publish time saved, cost saved, or any other productivity metric.

The caveat he put in writing. "Every change still goes through the same human review and security checks. It is eliminating tasks, not human judgment."

He closed by calling it an exciting time to push the frontier of productivity, "especially as we build the economy for agents (agentic finance)."

Disclosure: Not sponsored. No affiliate links. Not tested by MODEL. Armstrong's post describes his personal workflow, not a Coinbase-wide rollout.

MODEL Editors

Sources

  1. I've changed how I work over the last few weeks.

    X / @brian_armstrong

Keep reading

FREE · EVERY WEEK

The top AI tools this week, tested.

Every week MODEL's editors pick the AI tools worth your time, try them hands-on, and tell you what's good, what's not, and who it's for. Free, in your inbox.

Free. Unsubscribe anytime. Our picks are editorial and can't be bought.