Marketer Rachel Woods Gets 8 Hours of Autonomous AI Work Daily

AI strategist Rachel Woods says she gets about 8 hours of autonomous work per day from AI agents — not because she builds clever agents, but because she documents work in playbooks the AI must follow exactly. Woods, founder of AI Momentum Protocols, laid out her method on the AI Explored podcast.

Why does trust matter more than agent building?

Building an agent does not make AI autonomous. According to Woods, AI can only run unsupervised once it can be trusted to do the work the way it is supposed to be done. Without that trust, "an agent runs off doing random things, and someone has to go back and undo that work."

The fix is the playbook: a step-by-step process that AI follows to complete one specific type of work. When AI follows a playbook exactly, the result is predictable, and predictability is what lets owners stop checking the output. Woods said she has not found a type of work, across industries and roles, that cannot be playbooked. Her own email triage playbook runs every hour and keeps her at inbox zero.

How do you structure a playbook?

Every playbook has four components:

• Trigger — what kicks it off: a schedule, such as the hourly email run, or an event, such as a message or a new calendar invite.
• Inputs — what the AI needs; for email triage, that is entirely new emails.
• Steps — instructions written as if for a person, encoding the owner's decision-making and tagging.
• Output — the finished result; for the email playbook, a triaged inbox, drafted responses, and escalations of anything the AI cannot resolve.

Woods said the hardest part is setting the bookends. The common trap, familiar from delegating to people, is scoping too large — one playbook for a whole marketing campaign. A campaign includes briefs, emails, social posts, landing pages, follow-up lead magnets and multiple channels; one playbook for all of it will not earn autonomy until every piece works.

Her answer is one playbook per deliverable, tied together by an overall playbook. She calls this stacking. Her morning routine playbook triggers other playbooks and collects their outputs, a system "built brick by brick rather than overnight." Her tip: outline first, fill in detailed instructions later.

What is the CRAFT cycle?

CRAFT is Woods's five-step process for building and improving a playbook until AI runs it without review: clear picture, realistic design, AI-ify, feedback loop, team rollout. The cycle avoids the all-or-nothing trap of reviewing output forever or abandoning a playbook that stumbles.

The first version keeps a human in the loop; a second cycle on the same playbook, rebuilt from what was learned, is what removes the human. On feedback — which Woods called the most important step — she stressed that AI is predictive and guesses unless told otherwise. If the AI adds a useful unrequested summary, add "include a summary at the beginning" to the playbook; if not, instruct it never to. Claude can edit its own skills, but Woods's team constrains that with a major-and-minor framework: minor changes the AI makes alone, major ones get discussed first. Team rollout, the most overlooked step, is where ROI comes from; her tactic is habit stacking — defining exactly when each playbook runs in the broader workflow, plus training on starting and reviewing it.

Where should playbooks live?

Woods runs playbooks on ChatGPT, Claude and Grok; her students use Microsoft Copilot, and Gemini and Airtable also work. Her principle: "own the playbook, rent the tech." Playbooks stored in Google Docs, Notion or GitHub remain independent assets runnable in any tool — unlike keeping them only inside an AI tool, which she compared to a team member keeping important work in a personal drive. Her team uses Google Drive plus GitHub for cross-account collaboration.

Can AI manage other AI?

An AI middle manager oversees other agents and their playbooks. Woods uses two coordination methods: Claude subagents that run playbooks and report back, or a task tool such as Asana or ClickUp, where a manager agent assigns work and other agents respond to activity in the tool. Teams benefit from the audit trail of the task-tool method, which is why her company uses Asana; for personal routines she simply talks to her main agent, named Pepper. Woods called this "a clear and direct path to having the equivalent of a whole role, or even a whole team, doing work underneath one person" — and the management layer itself can be playbooked, repeating a level up.