Teammates learn from your answers

Your teammates were collecting everything you told them and using almost none of it. An answer about one lead was treated as too rare to trust, so the same question came back again. And one proposal left waiting could freeze a teammate’s learning for months. This release fixes both ends — what gets learned, and what gets applied — and puts you in charge of the balance.

Your answers count now

• A fact you state once stops coming back as a question — when you answer a teammate’s question, that answer becomes something it can learn from on its own. It no longer has to be repeated before it counts.
• Short answers are understood — a one-word reply like “yes” or “$500” is read alongside the question that produced it, so it is never mis-learned.
• One-off answers stay one-off — an answer that only applies to a single contact is recognized as such and is not turned into a standing rule.
• Conflicts name the page — when a newly learned fact contradicts something on a Knowledge page, the summary tells you which page.

Safe lessons apply, consequential ones ask

• Every lesson is classified by what it changes — a habit, a tone, a fact, or a new power. A wrong habit is one undo away; a wrong price waits for your confirmation.
• The safe half goes live right away — the rest arrives as a short “N to confirm” banner you approve, edit, or reject.
• Price changes read as a swap — when a learned figure replaces an old one, you see the old line struck through next to the new one, and the old line stays until you confirm.
• The learning dial — Settings → Teammates carries the team default: Ask me, Apply safe lessons, or Apply everything. Each teammate can override it. Giving an autonomous teammate full auto-apply takes an explicit confirmation.
• Learning keeps moving — a pending proposal no longer blocks the next round. A newer proposal replaces the older one, and approving it merges onto whatever your teammate has learned since.