Sherlock Suggests: Three Smart Questions to Get Every Conversation Started 🕵️
The biggest bottleneck to Sherlock adoption was never the AI, it was the blank box. Attendees and exhibitors open the chat, don't know what to ask, and give up before they even start. Smart Predefined Questions closes that gap: the moment someone opens Sherlock, three suggested questions are already waiting, tailored to that exact person and event.
How it works
- Sherlock's recommender engine pulls from five signals: the user's profile data, their past Sherlock interactions, the event's trend questions, the event timeline, and a set of hard-coded questions.
- It ranks and narrows those down to the three most relevant questions for that specific attendee or exhibitor, and surfaces them the moment they open Sherlock.
- Tap a suggestion, and it's sent straight to Sherlock as a message, kicking off a normal Sherlock reply, plan mode included, when the question calls for it.
Key behaviors
- Personalized when there's data: Returning users get questions ranked against their profile, their Sherlock history, trend questions, and the event timeline.
- Cold start, covered: New users with no profile data or interaction history still get three relevant questions, pulled from trend questions and a hard-coded set.
- No awkward gaps: Fewer than three good candidates? Sherlock shows whatever's available, no empty placeholders, no error states.
- Fallback built in: If the recommender comes up empty, or has an off moment, the hard-coded questions step in by default.
- No repeats: Once a question's been suggested and answered, or it's one Sherlock can't answer, it drops from that person's pool for the rest of the event.
- Always current: Close and reopen Sherlock, and the three questions recompute against the latest data available.
Why organizers will love it
Adoption for an AI agent lives or dies on that first message. Smart Predefined Questions turns Sherlock from "type something and hope" into a guided nudge, personalized per attendee, per exhibitor, per moment in the event. More people start a conversation, more of those conversations are relevant from message one, and matchmaking, session discovery, and exhibitor Q&A all get a running start instead of a cold open.