Hivework

Hivework is a platform for creating and operating AI agents that handle recurring work requiring judgment, not just fixed-rule tasks. Traditional workflow automation stops when a decision is needed, but Hivework fills this gap. Agents read context from existing tools, make decisions, and ask a human if unsure.

Agents are created on a visual canvas or by describing the job in plain language. No code is needed. Triggers include schedules, inbound email, webhooks, and manual starts, with conditions controlling whether a run fires.

Human approval is the default. Any action that sends, posts, or changes a record pauses for human approval, edit, or decline. Any unclassified action is treated as a write. Exceptions are narrow: one agent, one tool, revocable, and recorded.

Every run leaves a record of what the agent did, produced, cost, and declined to do. Runs survive a restart and resume from the last good step. Credentials are encrypted and never returned to the browser. Permission scopes are fixed.

Hivework is designed for operations owners at small and mid-sized companies needing internal automation. It's not suitable for engineers who prefer scripting or teams wanting fully autonomous automation without human approval.

The free tier includes a monthly allowance of runs with no card required; paid tiers increase the allowance.