‹ Recruiting & Onboarding recipesRecruiting & Onboarding · Agent Blueprint
Pre-debrief
Per-interviewer stance across the loop. How each interviewer landed, and where the panel disagrees, before the debrief.
Side by side: token usage, with and without Paperbase
~56k tokens saved · 70% less work for the agentWithout Paperbase
~80k tokens
- Collect each interviewer's notes after the loop.
- Skim them to reconstruct each person's stance.
- Recall who was enthusiastic and who had concerns.
- Guess at the disagreement before the meeting.
- Chair the debrief blind to where the panel stands.
With Paperbase
~24k tokens
- Pull each interviewer's stance from their notes and interactions.
- Group into hire, no-hire, and on-the-fence.
- Surface the exact point of disagreement, quoted.
- Prepare the debrief agenda around the contested points.
Agent prompt
You are preparing the debrief for {{candidate_name}}'s loop.
Using Paperbase:
1. Pull each interviewer's Interactions and extracted stance for the candidate.
2. Group stances: clear hire, clear no-hire, and hold.
3. For every disagreement, quote the evidence behind each side.
4. Note where a concern could be resolved with a reference or a follow-up.
Output:
- The panel's stance map.
- The points of disagreement, with the quote behind each.
- The questions the debrief must resolve to reach a decision.Placeholders in {double_braces} are inputs the agent will ask for at runtime. Give the prompt to any agent connected to Paperbase, and the rest grounds in your own memory graph.