Why this exists

The idea

Most outbound tools are built around throughput: sequences, cadences, thousands of sends, and a dashboard measuring how much left the building. They work, in the sense that email leaves. They also produce the mail everyone deletes.

StarlineSales assumes you are sending tens of emails a week, not tens of thousands, and that the bottleneck is writing something a person will actually reply to. So the AI writes a first draft and you edit it — and the product will not let a campaign go out with a single unread message in it.

Two decisions that shaped everything

Your own keys, never our markup

Live Search and AI Research run entirely through your own connected provider keys — your own Serper account, your own OpenAI account. We never see a query cost or a token bill, and we never mark one up. What we do store — the companies and prospects your searches actually find — is stored on the server, because a real search costs real money and cannot be safely left to a browser cache. It is yours: exportable and deletable on request, and never sold, shared, or used to train anything.

Nothing sends without a human

Every draft is read and approved by a person before it can be released. Not a setting, not a default that can be switched off — the campaign refuses to release otherwise. If you want a tool that sends AI-written email unattended, this is the wrong one, and that is deliberate.

Where it is

StarlineSales is in a free, open beta. It does the whole loop — import, qualify, draft, review, send, track — and there are things it does not do yet, listed plainly on the home page and the security page.

How we build

No trackers, no third-party scripts, no advertising pixels. The product loads code from its own origin and nowhere else. Decisions that shaped the architecture are written down as they are made, including the ones that turned out to be wrong and were reversed.