The Superhuman Go beta transition walkthrough shows the new agent structure working in practice
The specialised agents replacing what were previously separate features, Proofreader, Humanizer and Expert Review each as distinct agents with their own capabilities and personas, is the architecture that makes the platform feel more like a writing team than a writing checker.
The Humanizer agent's different personas, The Everyday Voice, The Professional Voice, others, are the register options that change what "humanise this" produces depending on the context. A single humanise button producing different outputs for different personas is more useful than a binary formal-informal toggle.
The Expert Review being positioned as access to human professional expertise within the AI-first platform is worth evaluating on the turnaround time and quality relative to hiring a human editor independently. If the quality is comparable and the turnaround is faster, it changes the editing workflow for professional writers significantly.
The context-aware suggestions based on the text field being used are what make the browser extension genuinely ambient rather than requiring you to visit a separate editing environment.
Have you tried Expert Review and how does the feedback compare to working with a human editor directly?