Questions and approvals

Tyms is built human-in-the-loop: your assistant works autonomously, but it checks in with you at the two moments that matter — when your request is ambiguous, and before actions that need your consent.

Clarifying questions

When a request could reasonably go more than one way, your assistant pauses and asks — rather than guessing and producing the wrong thing.

• Questions are structured: you’ll usually get options to pick from, plus a free-text box if none of them fit.
• You can skip the questions, and your assistant will proceed with sensible defaults.
• Pending questions wait for you. They survive page reloads and show up in other open tabs, so you can answer whenever you’re back.

Answering well here is the highest-leverage moment in the whole task — a ten-second answer often saves a full rework.

Approvals

Some tools your assistant can use perform real actions in connected systems. Before using one of those, your assistant posts an approval request right in the conversation, showing what it wants to do. You approve or deny it inline, and the task continues accordingly.

Two things keep this from becoming click-fatigue:

• Read-only tools are auto-approved by default — looking things up doesn’t interrupt you.
• Admins set the policy. For each connected app, workspace admins choose which tools always need approval and which don’t.

Next steps

• Apps in chat — the interactive apps your assistant can use.