Pipelines
Work is a flow, not a flat list.
Most task managers treat work as items to check off. Workers treats work as a flow: stages with clear ownership, each stage with its own agent, its own checks, and its own rules for moving forward.
- Multi-stage pipelines, each stage owned and observable
- Automated checks verify before anything moves on
- Rules you define — what advances, what waits
- Full audit trail of every action taken
Gates
The human stays in the loop — by design.
Not every decision should be automated. Gates are checkpoints where a person reviews, approves, or redirects. The agent does the work; you make the calls. Every gate decision is logged and auditable — nothing about the work is a black box.
- Approval workflows you configure per stage
- Automatic escalation when the agent is uncertain
- Notifications where you already are — Slack, email, in-app
- Every decision logged, every change attributed
Coding agents
Describe it in Slack. Get a real pull request.
Coding work flows through the same pipes. Describe what you need in a public channel — an agent spins up in a sandbox, writes the code, runs the tests, and opens a pull request in the thread. CI verifies before anything reaches a branch.
- Slack-native: public channels, thread replies
- Sandboxed: isolated containers, your repos stay safe
- CI loop: lint and test before any code moves
- Bring your own model keys, or use ours
Workers is coming soon.
The work layer is being built on the same beliefs and the same reading layer. Mark is live today.