‹ People & HR recipesPeople & HR · Agent Blueprint
Calibration season
Draft a performance review from the work itself. What this person actually owned, decided, and shipped - not what their self-review claims.
Side by side: token usage, with and without Paperbase
~123k tokens saved · 82% less work for the agentWithout Paperbase
~150k tokens
- Ask each manager to recall what their reports did all year.
- Skim project docs and tickets attributed to the person.
- Read self-reviews and try to verify the claims.
- Cross-check against PRs, launch notes, and meeting mentions.
- Reconstruct contribution from memory, missing the work no one wrote down.
With Paperbase
~27k tokens
- Pull the person's owned Work over the period: decisions, deliverables, shipped items.
- Attribute their contribution to cross-team outcomes.
- Pull their Decisions with rationale and impact.
- Compose the review draft from the actual record.
Agent prompt
You are drafting a performance review for {{person_name}} for {{period}}.
Using Paperbase:
1. Pull what {{person_name}} owned over the period: decisions, deliverables, shipped work, commitments met.
2. Pull their contributions to cross-team outcomes, attributed.
3. Surface the moments they led, pushed back, or took on the hard case.
4. Note any commitments missed or left open.
Output a review draft:
- What they actually shipped, with citations.
- Decisions they led and the impact.
- Strengths evidenced by the record, not the self-review.
- Growth areas: where commitments slipped, where scope was avoided.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.