How does the promotion system work from Training to Autonomous?
Every team starts in Training, where all proposed actions require your approval. As the team builds a track record of accepted proposals, it can be promoted to Supervised, where it earns trust by proposing work you accept. Finally, it can reach Autonomous, where it runs routine tasks independently but still leaves a receipt from the source system. Authority is granted per playbook rule, not as a blanket toggle, and each promotion is grounded in the team's actual performance.
What is a Play and how does it differ from a regular agent action?
A Play is a larger, multi-step workflow that a team proposes for approval. It includes a goal, the steps to achieve it, a credit budget, and a deadline. Once approved, the Play executes over days, with every step leaving a receipt. Regular agent actions are smaller, routine tasks like drafting a reply or scheduling a meeting. Plays are for bigger swings that require more coordination and resources.
How does YAGNI handle sensitive data like CRM records or financial transactions?
Connector credentials are customer-scoped, encrypted with AES-256-GCM, and resolved only at execution time; they are never placed in model prompts. The audit trail records the triggering user, tool, sanitized parameters, status, duration, and timestamp, but result payloads and message bodies are not copied into the audit row. For consequential actions like refunds, the agent stages a Decision that requires your explicit approval before execution.
Can I use YAGNI with my existing CRM, support desk, and code repository simultaneously?
Yes. YAGNI includes connectors for Slack, Gmail, Notion, Linear, and other tools. A single agent team can read and act across all connected systems. For example, a Sales Team can source leads from your CRM, draft replies in Gmail, and schedule meetings on your calendar, all from one identity with shared memory.
What happens if an agent team makes a mistake in Autonomous mode?
In Autonomous mode, routine and reversible actions execute on their own but always leave a receipt from the source system (e.g., a Stripe confirmation or a Gmail sent receipt). If a mistake occurs, you can review the receipt and adjust the team's playbook or demote its authority. Consequential actions (e.g., large refunds) still require your approval even in Autonomous mode.
How does the credit system work and what counts as a credit?
Credits are a monthly budget that covers all agent work: briefs, chat, drafts, and Plays. Connectors are included and do not consume credits separately. Each Play shows its credit budget before execution, so you know the cost upfront. Light daily reading uses fewer credits than heavy multi-step Plays. You can add credit packs at any time without changing your plan.
Can I start with a free trial and later scale to a larger team?
Yes. Every workspace starts free with 20 starter credits and no credit card required. You can connect your tools and let a team run real work. When you're ready to scale, you can choose a paid plan (Untethered for solo operators, Syndicate for small teams, Orbit for org-wide rollout, or Enterprise for custom needs). Credits are pooled across the workspace, so you can add team members without separate billing.
What is the Brief and how is it different from a dashboard?
The Brief is a daily, one-page artifact that composes the state of the entire business from all connected tools. It decides what matters, surfaces it, and carries a draft for the part you'd act on. Unlike a dashboard that waits for you to look, the Brief is proactively generated by the agent. You don't query for status; the agent reads the week and makes the call, showing you what it did and what's waiting on your decision.
