Dash demos by team
Approvals and guardrails
Yes. Dash asks for approval before it writes to any system, cites the source behind an answer, and says so when it cannot verify something rather than guessing. These demos show that behavior directly instead of asserting it.
This is the category that matters most before you connect an AI teammate to real tools. The question is not whether an agent can draft a message, it is what the agent does at the edges: when the answer is not in the data, when the doc and the channel disagree, and when a task is scheduled to run with nobody watching.
Each demo below is a conversation where Dash declines, escalates, or gates rather than proceeding.
Get started for free Chat with Dash, the AI teammate that works inside Slack and Microsoft Teams.
#ops Schedule an unattended task with approval gates Asked to schedule a write action, Dash separates approval consent from tool permission — and won't hand-wave the difference. View the conversation
#team Resolve a doc and Slack channel conflict Dash cites its source, notices a newer message contradicts it, and explains which one it trusts and why. View the conversation
#support Get a source, not a guessed number Asked for a number it can't verify, Dash says so — and offers a real way to get the answer instead of inventing one. View the conversation