Built around the thing that actually goes wrong.

The failure that matters in LinkedIn outbound is not a campaign that underperforms. It is a rep losing the account their whole pipeline lives in, usually after a fortnight of warning signs nobody was watching for. Rebuilding a network takes months and no feature makes up for it.

So the caps came first and everything else was built to fit inside them. Sending is checked against the limiter twice — once when the day is planned and again in the second before each message leaves — because minutes pass in between and the situation changes. Account health is polled nightly and read from every response. The list is deduplicated across the whole team so two reps never arrive in the same inbox.

The second decision was that judgement stays yours and typing does not. Three things need you — the strategy, the words, the launch — and each is a gate in the code rather than a promise in a paragraph. Everything after them can run unattended.

How much it finishes on its own is your call, per workspace. Hold pricing, legal and anything negative for yourself, or let it answer and book. Two holds you cannot switch off, and they are not the same kind of thing: a message the agent did not understand, because a reply written from a misread one reaches a real person just as fast as a good one; and a prospect who asks to speak to a human, because that one is not our caution, it is their wish. Somebody asking to remove them stops everything, always.

We publish the numbers we enforce, and we do not publish results we have not measured. When design partners produce real ones, those will appear here with their names on them.