Testing Siri AI on macOS 27: Early Impressions

The core challenge lies in the divide between Apple’s walled garden and the varied third-party workflows typical of professional laptop use. When asked to parse data, Siri AI successfully averaged benchmark scores from local screenshots, though it occasionally faltered by misinterpreting column labels or mixing synthetic and time-based results. These inconsistencies mean that, for now, manual verification remains a necessity for anyone relying on precise data.

Automation efforts also highlighted current limitations. Attempting to use Apple Intelligence to streamline benchmark testing via Shortcuts proved largely unproductive; the system struggled to execute actual software tasks, often defaulting to passive screen-capturing rather than triggering the applications themselves. Similarly, while Siri excels at surfacing images within Apple Photos or Messages, it remains largely blind to assets stored in third-party hubs like Lightroom Classic or Signal, regardless of whether the files reside locally.

Beyond data management, the assistant’s visual capabilities mirror the constrained utility of similar tools like Copilot Vision. It offers helpful guidance for Lightroom adjustments, yet its tendency to provide overly agreeable feedback—or occasionally suggest redundant settings—reveals a lack of critical nuance. While this iteration represents the most capable version of Siri to date, its effectiveness currently hinges on how deeply a user is embedded within the Apple ecosystem. For power users jumping between diverse platforms, the gap between potential and performance remains substantial.