The instruction you keep retyping becomes a slash command the whole team runs from the chat composer. Give it a cadence and it becomes a schedule: an agent runs it on its own and posts the result into chat or writes it up as a page.
Somebody works out the wording that gets a genuinely useful answer. It stays in their scratch file. Everyone else gets the mediocre version, and the recurring work — the Monday digest, the competitor sweep — depends on that person remembering.
Same question, six phrasings, six different answers, and no way to keep the best one.
The weekly summary happens when someone remembers it’s Friday.
Automation you can’t inspect is automation you stop trusting.
Write it once on the Commands page. It joins the “/” menu in the chat composer for everyone.
Type “/”, pick it, send. The whole team gets the good version of the prompt instead of their own approximation.
Weekdays, daily, weekly, hourly, or every N minutes. Pick which agent runs it and where the result lands — team chat, or a Docs page.
The Agents marketplace has a Prompts shelf; installing one adds it to the “/” menu, editable afterwards like any other.
The whole team sees every schedule and the history of each run. Creating, changing or pausing one is admin-only, because a schedule spends credits unattended.
Runs happen while someone has Decisive open, so a nightly job is waiting the next morning rather than firing at exactly 3am. Great for digests, wrong for minute-by-minute monitoring.
Three consecutive failures and a schedule stops, instead of burning credits in a loop.
Free for up to 10 people, with 1,000 AI credits included. No credit card, no sales call, no onboarding deck.