Skip to main content

On This Page

How a Solo Operator Runs a Full Company with Five Automation Primitives

3 min read
Share

These articles are AI-generated summaries. Please check the original sources for full details.

The five primitives I run a whole company on

The Solo Operator runs a consumer product company—physical inventory, two storefronts, multiple marketplaces, and a codebase—with a headcount of one. This is achieved using browser-operating agents that drive real sessions behind login walls, covering roughly two to four full-time roles’ worth of routine work.

Why This Matters

The technical reality is that 90% of small-business operations are locked behind login walls with no usable API, not the idealized 10% that already has one. This operator’s failures—such as a scraper returning stale data for nine days or a deploy shipping wrong subscription tiers—are survivable due to strict approval gates, not because agents are reliable. Anyone selling the version where agents don’t fail is selling something else.

Key Insights

  • A browser-operating agent drives real browser sessions with real logins to access seller dashboards, banking portals, and government sites; roughly 90% of small-business operations live behind a login wall with no usable API.
  • Persistent memory is structured per-site, per-project, and per-decision notes that agents read and write, preventing re-explanation and re-mistakes; the trap is dumping everything into one giant note, which worsens retrieval.
  • Approval gates convert a hundred hours of doing into one hour of reviewing, with agents drafting and the operator approving before irreversible actions like sending money, filing applications, or deploying to production.
  • Scheduled routines run recurring jobs like scanning grant portals daily, reporting ‘nothing changed’ explicitly to avoid silent breakage; the real unlock is closing open loops that occupy psychological attention.
  • The tooling costs a few hundred dollars a month, less than one day of one employee; failures are survivable because of gates, not because agents are reliable—e.g., a scraper returned stale competitor data for nine days without complaining.

Practical Applications

  • Use case: Browser agent for automating seller console and tax portal workflows where no official API exists; pitfall: skipping this primitive limits automation to only the 10% with APIs, leaving the operator doing the other 90% by hand at 11pm.
  • Use case: Coding agent for building storefronts, internal dashboards, and glue scripts, treated like a contractor with continuous probation; pitfall: the operator must read diffs to verify changes are contained, not write the features themselves.
  • Use case: Approval gates on money paths, such as payment processing and government filings, before automation; pitfall: scheduling first without manual verification leads to a cron job that ‘produces confident garbage every morning at seven.’

References:

Continue reading

Next article

Why I Replaced ESLint and Prettier with Biome: A Rust-Powered Alternative

Related Content