What we will not do in your place.

Laura is your sales intelligence copilot. She gathers, she cross-checks, she writes. She has no write path into your mailbox: when she suggests a message, she returns a proposal, and it is your own action that saves the draft. This is not a cautious setting someone could loosen — it is the absence of code.

• 1

  No tool of Laura's can send.

  The step that prepares a message returns a proposal. There is no send call behind it, and therefore nothing to switch off.

• 2

  The body of your messages never reaches the model.

  To draft a follow-up, Laura receives the counterpart, the subject line, and how long the conversation has been silent. Not the text of your exchanges.

Only one family of automations reaches a client without a salesperson having clicked just before: follow-up reminders, and only if your workspace has armed them. Six safeguards hold them. No setting turns them off — the function that decides whether a follow-up may go out receives only the target's situation and the current time. There is no checkbox to bypass one.

• 1

  A reply closes the sequence.

  The moment your counterpart replies, the follow-up stops — for good. It is the most important of the six: chasing someone who has just replied is the fastest way to be ignored.

• 2

  One follow-up, never a cadence.

  One automatic message per target and per rule, at most. No escalation, no five-step sequence, no copying in the manager on the third round. Whatever is still unanswered after one follow-up goes back to being a human matter.

• 3

  Opt-out is re-read on every send.

  Never from a cache. An opt-out recorded yesterday applies to today's message, and the check is redone target by target.

• 4

  Quiet hours.

  Nothing before 8:00, nothing after 19:00, nothing at the weekend. An automatic sales email arriving on a Sunday in the middle of the night is noticed — and it tells its recipient that no human read it first.

• 5

  A cap per workspace, per day.

  25 automatic follow-ups at most. A loop that runs away stops on a number, not on somebody's vigilance on a Friday evening.

• 6

  Nothing is armed by default.

  Every automatic follow-up stays off until a human in your workspace turns it on, and a single action turns everything off again.

The six safeguards answer the question "can this target be chased?". They do not answer "can this particular text go to this particular client today?". That is a different question, and a human answers it.

• 1

  What you read is what goes out.

  Subject and body are frozen at the moment the request is put to you. Nothing is recomposed at approval time: approving one text and sending another would be worse than showing nothing.

• 2

  A refusal holds for good.

  For that target, the follow-up you refused will not be put to you again the next day. Otherwise the only way out would be to disarm the whole automation.

• 3

  An undecided request expires.

  After 3 days it does not go out. A follow-up prepared last week speaks of a silence that may no longer exist.

A page explaining what we refuse to do is judged first on what it allows itself to write.

• 1

  No market figures.

  We were offered unsubscribe and sender-reputation statistics to back this page up. We could not tie them to a public source, named and dated. So they are not here: the only numbers written on this page are constants of the product.

• 2

  No delivery dates.

  What this page describes works today. What does not exist yet is not announced, not even approximately.

• 3

  No competitor names.

  We name capabilities — your mailbox, your CRM, your calendar — never other people's tools, and never to compare ourselves to them.