Stamp

Stamp's pitch is that it behaves like a secretary, not an inbox add-on, and the product design backs that up. The three learning features work together: Memories accumulate automatically about you and your contacts after every interaction, AI Labels let you write plain-English rules and specify the actions Stamp should take on anything filed under that label, and Context Search reads your inbox and calendar to understand your writing style and your availability before it writes or schedules. That combination is what makes a draft read like you wrote it, and it's the part a generic AI writer can't replicate — a generic writer has no memory of how you handled the same client three months ago. Stamp Mode is the workflow that makes this bearable at volume: instead of reading AI prose you didn't ask for, you approve or discard with one keystroke and keep moving. The agentic chat extends the same assistant to bulk drafting, web lookups, and arbitrary in-inbox actions, with automatic recipient population from your contacts and past threads. On mobile, the iOS and Android app plus Voice Mode lets you clear drafts and delegate hands-free. Where it gets less comfortable is the metering. Every tier above Free is defined by a multiplier — 10x usage on Personal, 25x on Business — and the caps are expressed in AI feature usage rather than in emails or drafts, so you can't easily predict from your inbox volume whether you'll fit. Connected accounts are also rationed, 2 on Personal and 5 on Business, and monthly profile enrichments are 50 and 500 respectively. Pricing is published and transparent: Personal is $20/month month-to-month with yearly billing saving 20%, Business is $50/month month-to-month with the same yearly discount, Enterprise is contact-only and adds support for any email provider plus on-prem deployments. A 30-day free trial exists, so the honest move is to run the trial on Free-tier mechanics and see whether your real volume brushes the cap before you commit to a paid seat. Limitations worth naming: no source this run names the underlying AI model or a public API, so treat model choice and programmatic access as unverified rather than absent. Enterprises needing compliance controls below the Enterprise tier will find nothing documented. And Stamp only makes sense if you're willing to let an AI read and remember your mail — the value and the cost are the same coin. Compared with Superhuman, you're trading a polished traditional client for an assistant that does the work; compared with Gmail's built-in Gemini features, you're paying for persistence and taste rather than convenience you already have.