A booking comes through a web form and someone ticks the small team bracket. Then in their first message they mention opening a second office next month and wanting everyone on the account. Two answers to the same question, both given in good faith.
When Rosebud manages the enquiry, the model picks the answers out of the conversation as people give them, and the business's own rules decide what happens next. When those answers do not add up, it does not guess.
Claude Opus 5.5 is ridiculously good at replying to enquiries by following your exact rules and writing like you do.
But most people are still using it like a better ChatGPT.
That's a waste ๐
With the right setup, it can handle the heavy lifting between the initial enquiry and the booked call.
And you can try it on your own enquiries.
Get free access / start your trial: https://t.co/5zFHNhTTi3
That holds at any size. One person looking after their own enquiries or a team handling thousands, the effort it took to win a booking should never be the thing that leaves it unprotected.
The booking is where the work changes, not where it stops.
A long conversation that ends in a booking feels like the finish line. In my opinion it is closer to halfway.
Think about the enquiries that take the most work to get booked.
Spend the conversation budget on getting them through the door, but never let that budget starve the reminder that keeps them there. Winning the job was never the whole job.