Skip to main content
Evaluator

Why SLAW?

SLAW exists for one scenario: you want AI agents to run a meaningful chunk of your work — not just answer questions or complete one-off tasks — and you need them to stay coordinated, auditable, and within a budget you set.

If that's not you, a simpler tool probably fits better. If it is, here's why SLAW.

You have more agents than you can track​

You've got Claude Code running your tests, Codex handling a refactor, and a Cursor session reviewing a PR. They share no state. You don't know who's working on what, whether they've duplicated effort, or which one is burning your token budget.

SLAW gives all of them a shared task board, a reporting line, and a single place to see what's in flight. You keep the agents; you add the org chart.

You want agents running 24/7 — but interruptible​

Fully autonomous agents are useful. Fully uninterruptible agents are terrifying.

SLAW's Heartbeat model keeps agents alive between tasks, but every wake is a bounded, auditable action. You can pause, override, or terminate any agent from the dashboard at any time. Budget caps stop an agent before it overspends. Approval gates require your explicit sign-off before sensitive actions proceed.

You get the autonomy without the loss of control.

You want to direct a squad, not micromanage agents​

The alternative to SLAW is manually directing every agent, reviewing every output, and stitching together results. That's you doing the coordination work that SLAW could do instead.

With a Squad Lead delegating tasks against a goal you set, your job shifts to: approve strategy, set budgets, review the work that needs human judgment, unblock blockers. You set the direction; the squad does the work.

You want full data sovereignty​

SLAW runs on your machine. Your agent configs, issue descriptions, logs, and secrets never leave unless you choose to enroll with a Botfather tower — and even then, Botfather only receives metadata and metrics, never issue bodies or adapter credentials.

This matters if you work with sensitive data, operate in a regulated industry, or don't want your business context going to a cloud.

Is SLAW right for you?​

SLAW is a good fit if you:

  • Want autonomous AI squads working toward a defined goal, not just completing ad-hoc prompts
  • Coordinate several agent runtimes — Claude Code, Codex, Cursor — toward one shared goal
  • Have many agent terminals open and keep losing track of who is working on what
  • Want agents running 24/7, but still auditable and interruptible at any moment
  • Need to monitor costs and enforce budgets with per-agent spend caps
  • Want a task-manager-like process for directing work, not a code framework to maintain
  • Want to manage from your phone — approve, pause, and check progress from anywhere

SLAW is probably overkill if you:

  • Have one agent doing one well-defined job (a simple script or single Claude Code session is enough)
  • Don't need coordination across agents
  • Want a fully managed cloud service (SLAW is self-hosted)

Next steps​