Key takeaways
- Explicit rules are fast, predictable, and private. Always-work domains, distracting defaults, subdomain exceptions, and time limits should handle the common cases before any model evaluates context.
- When a site is ambiguous, ask a short question: Is this work-related? The answer can apply once, for the session, or become a saved rule.
- Personalization should reduce repeated questions while preserving control. Users need to see why a site was challenged and undo incorrect classifications immediately.
Use rules before AI
Explicit rules are fast, predictable, and private. Always-work domains, distracting defaults, subdomain exceptions, and time limits should handle the common cases before any model evaluates context.
Ask when confidence is low
When a site is ambiguous, ask a short question: Is this work-related? The answer can apply once, for the session, or become a saved rule.
Learn without becoming opaque
Personalization should reduce repeated questions while preserving control. Users need to see why a site was challenged and undo incorrect classifications immediately.
Frequently asked questions
What is the main idea behind context-aware blocking for sites that are sometimes work?
Handle subdomains, work accounts, research paths, and ambiguous tools without disabling protection.
How does Dialed approach this?
Dialed uses adjustable browser boundaries, explicit context, and transparent behavioral signals rather than treating every site or workday the same.