It uses the public API, as you do
Every question runs through the same API a script or an external client uses, on a token minted for that turn and revoked after it. The assistant can reach nothing you cannot reach yourself.
AI assistant
One question in plain language, and the answer cites the contacts, lists and campaigns behind it. It reads everything you can read, and it never writes without asking first.
Every question runs through the same API a script or an external client uses, on a token minted for that turn and revoked after it. The assistant can reach nothing you cannot reach yourself.
Looking something up needs no confirmation. Anything that changes your workspace pauses on a card naming exactly what it is about to do; declining answers the question instead.
Each step appears as it runs, because a real answer can take tens of seconds — and a panel that only shows a spinner leaves you wondering whether it is stuck.
Every thread stays in a history you can reopen. A failed question keeps its wording, so asking again costs no typing — and no quota.
Smart
It can queue a message the way you would, and it will ask you first. Nothing leaves your LinkedIn account without a confirmation, and the daily limits and office hours you set apply to it exactly as they apply to you.
Keep reading
An AI assistant with read access answers questions no dashboard was built for. One with write access needs a rule about when it stops. Here is what that looks like in practice.
Swimlane ships an MCP server: point Claude, Cursor or the OpenAI API at your workspace and ask for the work instead of writing the calls. Setup, five real examples, and the guardrails.
Swimlane's public API covers the whole loop: import contacts from LinkedIn, work them in the CRM, build and run campaigns. A tour with the calls that matter and the traps worth knowing.
Connect LinkedIn in two minutes. Free to start, and free to leave with your data.
7 days free · Cancel anytime