Floom

Floom is a tool for setting up AI workers to handle recurring tasks in a loop. Its page describes examples such as summarizing Gmail and posting to Slack, triaging new pull requests into a review list, and drafting follow-up emails when a new row appears in a spreadsheet. The product is aimed at teams that turn repeated work into loops and want an AI worker to run on a schedule or in response to a trigger.

It works through agent integrations and a command-line interface. Floom can be added to Claude Code or Cursor, and it also says it runs from a terminal with a CLI and a REST API. A loop is set up by connecting Floom to an agent, choosing a worker, giving it a cadence or trigger, and approving it once. The page also shows a prompt-based onboarding flow and a manual install command, npx @floomhq/floom mcp install.

Several named workers and templates are listed, including Inbox Cleaner, Follow-up Drafter, Candidate Screener, Content Publisher, Lead Enricher, and Shipped Digest. The product says each loop runs quietly in the background, asks before it acts on sensitive steps, and keeps every run on the record with full logs, inputs, and outputs. It also says loops can be stopped, edited, or retired at any time. The site states that hosted workers are available now and that users can build their own loops soon.

It identifies Floom as built in San Francisco and includes links for GitHub, product information, pricing, templates, integrations, docs, company, about, security, legal, privacy, and terms.