# How much should an AI decide without asking you first?

DevFeed: [How much should an AI decide without asking you first?](<https://devfeed.tech/articles/how-much-should-an-ai-decide-without-asking-you-first-121947.md>)

Original publisher: [Read original article](<https://zapier.com/blog/how-much-should-ai-decide-without-asking>)

Author: Jay Ayinla

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [Zapier](<https://devfeed.tech/sources/zapier.md>)

Topics: [AI interaction best practices](<https://devfeed.tech/topics/ai-interaction-best-practices.md>), [User interface design](<https://devfeed.tech/topics/ui-design.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human centric design](<https://devfeed.tech/topics/human-centric-design.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [ai-tool](<https://devfeed.tech/tags/ai-tool.md>), [ai-workflows](<https://devfeed.tech/tags/ai-workflows.md>), [ai-workspace](<https://devfeed.tech/tags/ai-workspace.md>), [automation](<https://devfeed.tech/tags/automation.md>), [chat](<https://devfeed.tech/tags/chat.md>), [code](<https://devfeed.tech/tags/code.md>), [code-execution](<https://devfeed.tech/tags/code-execution.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [engineering](<https://devfeed.tech/tags/engineering.md>)

## AI overview

Sidekick's permission system evolved from asking before every change to applying clear rules to the resources at risk--such as files, connected apps, websites, and automations. "Never" rules take precedence, then "ask first"; uncovered actions may proceed under the selected mode. Prompts now explain the scope and rule involved, and reviews appear in a panel and an inbox rather than being buried in chat. The article describes Auto mode, where an AI reviewer evaluates uncovered actions and asks when unsure, and a pilot YOLO mode that allowed uncovered actions but still honored explicit restrictions and required approval to change permissions. YOLO was later slated for removal. A new mode is in progress: rules would stay constant across modes, while the mode determines how uncovered actions are handled. The planned approach would ask before actions reaching outside Sidekick and never allow legal agreements or payments. The authors' practical tests are whether users can state their rules plainly and understand which rule allowed or blocked an action.

## Source excerpt

Any AI product that does things, not just says things, runs into the same decision problem sooner or later. The AI assistant can ask before acting, or act and tell you afterward. Both easy answers are wrong. If it asks every time, you stop reading the questions and click yes on reflex, so the questions end up protecting nobody. If it never asks, everything is fine until the first time it does something you didn't expect, and that single surprise is enough to end your trust in it. We've been work