Two communities hit the same wall in the past two weeks: bloggers drowning in SEO/link-building solicitations (lobste.rs), and job seekers flooded by LLM-personalized recruiting spam harvested from HN hiring threads (969-point thread). Both report that this mail comes from properly authenticated servers and sails past spam filters, leaving crude keyword rules as the only defense. The want: something that classifies sender intent on top of an existing Gmail/IMAP inbox instead of requiring a new email provider.
Protocol-level signals are exhausted — the spam IS well-authenticated, so the classifier has to read intent (link-building, guest-post, templated recruiting) the way a human does; ship it as a Gmail filter/label layer first, because HEY already proved the screening model and also proved that 'switch your email provider' is the adoption killer.
landscape (3 existing solutions)
Email authentication standards authenticate the sending domain and pass SPF/DKIM/DMARC checks, which is enough to clear most spam filters — but authentication is not delivery, and providers can still apply reputation and content signals on top of it. The practical effect is that LLM-personalized outreach from real authenticated domains sails past content filters, leaving intent classification as the only remaining defense. LLMs made it cheap to personalize at scale; nothing consumer-grade classifies sender intent on top of an existing inbox, leaving a gap between switching providers entirely (HEY) and hand-written keyword filters.