AI Services Market › Use cases › Work & Operations
Work & OperationsDone with custom-built or unnamed agent 2026-10-29
Run a one-person company on AI agents
Over six months the poster ran a company with agents reading and writing the business as text, including turning one LinkedIn post into multiple pieces of content and managing engagement, which backfired with slop copy.
- What They Did
- The poster put the business into a single git repo as plain-text folders and playbooks, then used one generalist agent to route tasks across sales, CRM, content, outreach, and ops. Agents drafted work into one approval queue while browser and CLI tools handled LinkedIn, Instagram, Google Workspace, and internal apps; the human approved sends, reviewed edits, and handled irreversible actions.
- How It Went
- Over six months it closed a first paying client, found and closed two cold leads, replaced about ten subscriptions with owned tools, and reached five-figure revenue. It also misfired: social automation got an account suspended, drafts and queues stalled unnoticed, and AI-written engagement on repurposed content backfired as slop copy.
- Worth Knowing
- The poster says multi-agent runs cost about 15x the tokens, so they only use subagents for genuinely complex parallel work.
Try it yourself with custom-built or unnamed agent
I run [type of business] and write about it on LinkedIn. Take this post: [paste post]. Turn it into [number] pieces such as [e.g. a newsletter section, a short thread, a blog outline], written in my voice using these samples: [paste 2-3 of your past posts]. Avoid generic filler phrases and keep each piece specific to my experience. Give me drafts to review and do not publish or reply to anyone without showing me first.
Do this with a skill.
These AI Services Market playbooks do this exact job.