Your client's LinkedIn account is the asset
Losing one is a conversation you do not want to have. Cadence sends their invitations at a pace the account survives, to the people most likely to accept, and tells you the truth about every one that did not go out.
Every agency running LinkedIn outreach hits the same three walls. Most tools are built as though only the first one exists.
It happens in bursts, not in totals. An account can send a modest number of invitations in a day and still get flagged for sending them too close together. Then your client is locked out of their own profile and you are the one explaining it.
LinkedIn gives every account a small number of invitations a week. Spending one on somebody who has not opened LinkedIn since March is a slot gone. Most tools send in whatever order the list happened to arrive in.
The client asks which channel is producing meetings. You have a number of invitations sent and no way to trace an accepted connection back to how that person was found.
LinkedIn answers a refused invitation with a success code and a failure inside the body. Read only the code, as many tools do, and you report a send that never happened. Your client's reporting is then wrong in the flattering direction, which is the worst direction for it to be wrong in.
What that looks like in practice
When an account hits LinkedIn's own ceiling, Cadence says so in words, names the account, and says when it will try again. You never get a green number that means nothing. Read what we measured about why accounts get signed out.
A list of profiles, from wherever you found them. There is no limit worth worrying about, because nothing is sent in a burst anyway.
Spaced out, never on a fixed rhythm, and paused the moment LinkedIn signals it has had enough. Each client account has its own budget and its own pace, so a busy one never affects a quiet one.
Who went out, who accepted, who LinkedIn refused and why. Broken down by where each person came from, so the channel question has an answer.
You have a list of two hundred and about a dozen invitations a day. The order is the whole game.
Ask for a specific person and they go to the front, ahead of whatever is already queued. The answer is when they will go out, not a wait measured in weeks.
If somebody was found because they commented on a post last Tuesday, that is not a guess about whether they use LinkedIn. It is something we watched them do, and they get the slot.
Existing connections are recognised before anything is spent on them, so your scarce weekly invitations are never used discovering somebody was already in your network.
Not separate lists inside one queue. Separate budgets, separate pacing, separate agendas, and separate reporting.
Pacing and budgets
one client's volume never borrows from another's
Acceptance reporting
including which source produced it
Blocks and recovery
one restricted account never stops the others
LinkedIn accounts
running through this infrastructure today
There is no interface you have to adopt. Ask Claude to connect with a list of people and it goes onto the queue, paced, with the same answer you would get anywhere else.
Already using Claude?
Cadence is part of the BeReach connector, so it works inside Claude and Claude Cowork with nothing else to install. See the Claude connector.
Whatever LinkedIn allows that account, which is not a published number and is not the same for everyone. Cadence does not add a cap of its own on top. It sends until LinkedIn signals it has had enough, records exactly what LinkedIn said, and waits. Anyone quoting you a precise weekly figure is repeating a guess.
Nothing is written down for them at all, which is the point. The queue is recalculated every time rather than planned in advance, so when the account can send again they simply reappear, re-ranked against anyone who became more relevant meanwhile.
Sending stops for that account and only that account, and you are told which one and why. Their queue is kept intact, and it resumes when the account is reconnected. Nothing is lost and nothing is silently dropped.
That is the default. Anyone you name goes to the front, and asking again moves them back to the front. Ordering is for lists you built in advance, not for the person you just asked about.
No. It lists what is still pending from each account so you can see the backlog, and the decision stays yours. Pending invitations do count against your quota, so leaving them outstanding is a quiet tax on your own limit, but LinkedIn imposes a multi-week wall before you can re-invite somebody you withdrew, which makes it a call worth taking one at a time rather than on a schedule.
LinkedIn's terms restrict automated access, and that is worth reading before you buy any tool in this category, including this one. What Cadence changes is the exposure: it sends from your client's own account, at a human pace, and it stops on the first signal that LinkedIn is unhappy rather than pushing through it.
Point it at a list, watch a week of sending, and check the acceptance numbers against whatever you are using now. Add the rest when it has earned it.
Click a button and let your favorite AI weigh in.
